CN117689253A - Method, device, equipment and medium for generating flow information of product model - Google Patents
Method, device, equipment and medium for generating flow information of product model Download PDFInfo
- Publication number
- CN117689253A CN117689253A CN202311674499.XA CN202311674499A CN117689253A CN 117689253 A CN117689253 A CN 117689253A CN 202311674499 A CN202311674499 A CN 202311674499A CN 117689253 A CN117689253 A CN 117689253A
- Authority
- CN
- China
- Prior art keywords
- model
- historical
- information
- product
- model product
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 230000000694 effects Effects 0.000 claims abstract description 227
- 230000009466 transformation Effects 0.000 claims abstract description 33
- 238000012216 screening Methods 0.000 claims abstract description 16
- 230000008859 change Effects 0.000 claims description 36
- 239000011159 matrix material Substances 0.000 claims description 36
- 238000013499 data model Methods 0.000 claims description 31
- 230000007246 mechanism Effects 0.000 claims description 13
- 238000006243 chemical reaction Methods 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 9
- 230000004044 response Effects 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 4
- 230000002123 temporal effect Effects 0.000 claims description 4
- 238000003062 neural network model Methods 0.000 description 29
- 230000008569 process Effects 0.000 description 10
- 230000006870 function Effects 0.000 description 8
- 230000007423 decrease Effects 0.000 description 7
- 238000004891 communication Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 230000008901 benefit Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000009469 supplementation Effects 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 3
- 238000012163 sequencing technique Methods 0.000 description 3
- 238000010606 normalization Methods 0.000 description 2
- 238000012805 post-processing Methods 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 238000012512 characterization method Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Educational Administration (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the invention discloses a method, a device, equipment and a medium for generating flow information of a product model. One embodiment of the method comprises the following steps: acquiring a historical model product calling amount sequence, a historical model product using effect information sequence and model product information; determining call amount transformation trend information and determining effect transformation trend information; determining trend difference information; performing sequence division on the historical model product call quantity sequence and the historical model product use effect information sequence; screening historical model products to call quantum sequences; determining at least one first historical model product use effect information subsequence with a time corresponding relationship; generating data product model flow prediction information for a target future time; and putting the value putting information of the corresponding volume in the target client. According to the method and the device, the flow information of the product model can be quickly and accurately generated, so that the value of the target data product model can be accurately put in later.
Description
Technical Field
The embodiment of the disclosure relates to the technical field of computers, in particular to a method, a device, equipment and a medium for generating flow information of a product model.
Background
Currently, data model products are widely used in daily life of people. The flow of the data model product can effectively measure the popularity of the product. For the generation of flow information of a data model product, the following methods are generally adopted: the rough flow information of the data model product is determined manually and intuitively through various product indexes.
However, the inventors have found that when the above-described manner is adopted, there are often the following technical problems:
the product flow is predicted manually only through various product indexes intuitively, and the product model flow information generated is not accurate enough due to the fact that the product flow is large in one-sided performance, so that accurate value throwing cannot be performed on target data model products in the follow-up process.
Continuously, in the process of solving the technical problem that the flow information of the product model cannot be quickly and accurately generated by adopting the technical scheme so as to facilitate the follow-up accurate delivery of value aiming at the target data product model, how to generate flow prediction information at the corresponding future time aiming at the using effect information subsequence of the historical model product and the model product information is key technical content.
For the prediction of flow prediction information at future times, conventional solutions are generally: and directly inputting the historical model product calling quantum sequence and the model product information into a model flow prediction information generation model to obtain flow prediction information in future time. However, the above solution has the following technical problems:
The time period corresponding to the historical model product calling quantum sequence may not be continuous, so that the characteristic information reflected by the historical model product calling quantum sequence is also not continuous, and is directly input into a subsequent model flow prediction information generation model, so that the extracted characteristic of the model flow prediction information generation model is limited and discontinuous, and the flow prediction information at the future time is not accurate enough.
The above information disclosed in this background section is only for enhancement of understanding of the background of the inventive concept and, therefore, may contain information that does not form the prior art that is already known to those of ordinary skill in the art in this country.
Disclosure of Invention
The disclosure is in part intended to introduce concepts in a simplified form that are further described below in the detailed description. The disclosure is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter.
Some embodiments of the present disclosure propose a method, apparatus, device, and medium for generating product model flow information to address one or more of the technical problems mentioned in the background section above.
In a first aspect, some embodiments of the present disclosure provide a method for generating flow information of a product model, including: acquiring a historical model product calling amount sequence corresponding to a target data model product, a historical model product using effect information sequence and model product information; determining call amount change trend information corresponding to the call amount sequence of the historical model product and effect change trend information corresponding to the use effect information sequence of the historical model product; determining trend difference information between the call volume change trend information and the effect change trend information, wherein the difference time length corresponding to the trend difference information is longer than the target time length, and carrying out sequence division on the historical model product call volume sequence and the historical model product use effect information sequence according to the trend difference information to obtain a historical model product call quantum sequence set and a historical model product use effect information subsequence set, wherein the historical model product call quantum sequence set and the historical model product use effect information subsequence set have a one-to-one correspondence; screening historical model product calling quantum sequences corresponding to the first trend types from the historical model product calling quantum sequence set to obtain at least one first historical model product calling quantum sequence, and determining at least one first historical model product using effect information subsequence with a time corresponding relation with the at least one first historical model product calling quantum sequence; and generating data product model flow prediction information aiming at a target future time according to the data product model flow prediction information, and throwing value throwing information corresponding to the volume at a target client according to the data product model flow prediction information so as to display the value throwing information at the target client.
In a second aspect, some embodiments of the present disclosure provide a product model flow information generating apparatus, including: the acquisition unit is configured to acquire a historical model product calling amount sequence, a historical model product using effect information sequence and model product information corresponding to the target data model product; a first determining unit configured to determine call amount transformation trend information corresponding to the call amount sequence of the history model product, and determine effect transformation trend information corresponding to the effect information sequence of the history model product; a second determining unit configured to determine trend difference information between the call-up conversion trend information and the effect conversion trend information, wherein a difference time period corresponding to the trend difference information is longer than a target time period; the dividing unit is configured to divide the historical model product calling quantity sequence and the historical model product using effect information sequence according to the trend difference information to obtain a historical model product calling quantum sequence set and a historical model product using effect information subsequence set, wherein the historical model product calling quantum sequence set and the historical model product using effect information subsequence set have a one-to-one correspondence relation with each other in time, and the screening unit is configured to screen a historical model product calling quantum sequence corresponding to a first trend type from the historical model product calling quantum sequence set to obtain at least one first historical model product calling quantum sequence; a third determining unit configured to determine at least one first history model product use effect information subsequence in a temporal correspondence with the at least one first history model product invocation quantum sequence; a generation unit configured to generate data product model flow prediction information for a target future time based on the at least one first historical model product invocation quantum sequence, the at least one first historical model product usage effect information sub-sequence, and the model product information; and the delivery unit is configured to deliver the value delivery information of the corresponding volume to the target client according to the data product model flow prediction information so as to display the value delivery information on the target client.
In a third aspect, some embodiments of the present disclosure provide an electronic device comprising: one or more processors; a storage device having one or more programs stored thereon, which when executed by one or more processors, cause the one or more processors to implement the method as described in any of the implementations of the first aspect.
In a fourth aspect, some embodiments of the present disclosure provide a computer readable medium having a computer program stored thereon, wherein the program when executed by a processor implements a method as described in any of the implementations of the first aspect.
The above embodiments of the present disclosure have the following advantageous effects: the product model flow information is quickly and accurately generated by the product model flow information generation method of some embodiments of the present disclosure, so that the subsequent accurate delivery of value to the target data product model is facilitated. Specifically, the reason why the accurate delivery of value to the target data product model is not possible is: the product flow is predicted manually only through various product indexes intuitively, and the product model flow information generated is not accurate enough due to the fact that the product flow is large in one-sided performance, so that accurate value throwing cannot be performed on target data model products in the follow-up process. Based on this, in the product model flow information generating method of some embodiments of the present disclosure, first, a historical model product call amount sequence, a historical model product use effect information sequence and model product information corresponding to a target data model product are obtained for prediction of data product model flow prediction information at a subsequent target future time. And then determining call amount transformation trend information corresponding to the call amount sequence of the historical model product and determining effect transformation trend information corresponding to the using effect information sequence of the historical model product. Here, the calling amount change trend information and the effect change trend information are determined to determine the difference between the calling amount change trend information and the effect change trend information, so that prediction supplementation of corresponding trend information is performed according to the difference between the calling amount change trend information and the effect change trend information, and the situation that larger errors occur in the process of generating data product model flow prediction information in the follow-up process due to information errors in the history model calling amount sequence and the history model product use effect information sequence is avoided. And then, determining trend difference information between the call volume transformation trend information and the effect transformation trend information as an actual information sequence for predicting the model product call volume and the model product utilization effect information of the subsequent rest time period. Wherein, the difference time length corresponding to the trend difference information is longer than the target time length. And then, according to the trend difference information, carrying out sequence division on the historical model product calling quantity sequence and the historical model product using effect information sequence to obtain a historical model product calling quantum sequence set and a historical model product using effect information subsequence set, wherein the historical model product calling quantum sequence set and the historical model product using effect information subsequence set have a one-to-one correspondence. Here, through the sequence division, the call volume sequence and the usage effect information subsequence corresponding to the trend type are determined later. And screening out a historical model product calling quantum sequence corresponding to the first trend type from the historical model product calling quantum sequence set to obtain at least one first historical model product calling quantum sequence for generating flow prediction information of a subsequent data product model. And further determining at least one first historical model product use effect information subsequence in a time corresponding relation with the at least one first historical model product calling quantum sequence, so as to be used for generating flow prediction information of a subsequent data product model. Further, based on the at least one first historical model product invocation quantum sequence, the at least one first historical model product usage effect information sub-sequence, and the model product information, data product model flow prediction information for a target future time may be accurately generated. And finally, according to the data product model flow prediction information, value delivery information corresponding to the volume is delivered to a target client side, so that the value delivery information is displayed on the target client side, and the model related content of the data product model is displayed more widely.
Drawings
The above and other features, advantages, and aspects of embodiments of the present disclosure will become more apparent by reference to the following detailed description when taken in conjunction with the accompanying drawings. The same or similar reference numbers will be used throughout the drawings to refer to the same or like elements. It should be understood that the figures are schematic and that elements and components are not necessarily drawn to scale.
FIG. 1 is a flow chart of some embodiments of a product model flow information generation method according to the present disclosure;
FIG. 2 is a schematic structural view of some embodiments of a product model flow information generating apparatus according to the present disclosure;
fig. 3 is a schematic structural diagram of an electronic device suitable for use in implementing some embodiments of the present disclosure.
Detailed Description
Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete. It should be understood that the drawings and embodiments of the present disclosure are for illustration purposes only and are not intended to limit the scope of the present disclosure.
It should be noted that, for convenience of description, only the portions related to the present invention are shown in the drawings. Embodiments of the present disclosure and features of embodiments may be combined with each other without conflict.
It should be noted that the terms "first," "second," and the like in this disclosure are merely used to distinguish between different devices, modules, or units and are not used to define an order or interdependence of functions performed by the devices, modules, or units.
It should be noted that references to "one", "a plurality" and "a plurality" in this disclosure are intended to be illustrative rather than limiting, and those of ordinary skill in the art will appreciate that "one or more" is intended to be understood as "one or more" unless the context clearly indicates otherwise.
The names of messages or information interacted between the various devices in the embodiments of the present disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.
The present disclosure will be described in detail below with reference to the accompanying drawings in conjunction with embodiments.
Referring to fig. 1, a flow 100 of some embodiments of a product model flow information generation method according to the present disclosure is shown. The flow information generating method of the product model comprises the following steps:
And step 101, acquiring a historical model product calling amount sequence, a historical model product using effect information sequence and model product information corresponding to the target data model product.
In some embodiments, an execution body (for example, an electronic device) of the product model flow information generating method may acquire a historical model product call volume sequence, a historical model product use effect information sequence and model product information corresponding to the target data model product through a wired connection manner or a wireless connection manner. The target data model product can be a data model product in the target field. For example, the target area may be a credit field. The data model product may be a credit score data model in the credit field. In practice, the credit score data model may be a credit score model based on a time-series neural network model. The historical model product call volume sequence can be a call volume sequence which corresponds to the target historical time period and is called by the characterization model product history. For example, the target history period is "1 month 1 day-1 month 5 days". The corresponding historical model product calling amount sequence is { "1 month and 1 day: 300","1 month 2 days: 499","1 month 3 days: 284","1 month 4 days: 299","1 month and 5 days: 672"}. The historical model product use effect information sequence may be an effect information sequence corresponding to the target historical time period and representing a model effect represented by the model product history. In practice, model effect information may be embodied with various model verification indicators. For example, the actual effect of the model may be represented by the accuracy. For example, the target history period is "1 month 1 day-1 month 5 days". The corresponding historical model product calling amount sequence is { "1 month and 1 day: 0.4","1 month 2 days: 0.5","1 month 3 days: 0.2","1 month 4 days: 0.34","1 month 5 days: 0.5"}. The model product information may be product information corresponding to the data model product. For example, the model product information may include, but is not limited to, at least one of: model product identification, model product function information and model product advantage information.
It should be noted that, the time difference corresponding to each two adjacent historical model product call amounts in the historical model product call amount sequence is a predetermined time length.
Step 102, determining call volume transformation trend information corresponding to the call volume sequence of the historical model product, and determining effect transformation trend information corresponding to the using effect information sequence of the historical model product.
In some embodiments, the executing entity may determine call volume transformation trend information corresponding to the call volume sequence of the historical model product, and determine effect transformation trend information corresponding to the effect information sequence of the historical model product. The call volume change trend information can represent the change condition of the call volume under each time slice. The effect transformation trend information may characterize the transformation of the effect information under each time segment. For example, the effect transformation trend information may characterize the transformation of KS values at various time segments.
As an example, the execution body may subtract the previous call volume from the next call volume in the historical model product call volume sequence to obtain the subtracted value sequence. Then, the first value in the subtracted value sequence is set to 0, and the last value is set to be the same as the last-last value, so that a value sequence is obtained as call-in transformation trend information. The effect transformation trend information can be specifically generated by referring to the call-up transformation trend information.
Step 103, determining trend difference information between the call conversion trend information and the effect conversion trend information.
In some embodiments, the execution subject may determine trend difference information between the call conversion trend information and the effect conversion trend information. Wherein, the difference time length corresponding to the trend difference information is longer than the target time length. The trend difference information may characterize a trend difference between the call-in transformation trend information and the effect transformation trend information. The trend difference information includes a plurality of trend difference sub-information. Each trend difference sub-information corresponds to a time period longer than the target time period. The trend difference sub-information characterizes the difference of the transformation trend of the corresponding historical model product calling quantum sequence and the corresponding historical model product using effect information sub-sequence.
And 104, according to the trend difference information, carrying out sequence division on the historical model product calling quantity sequence and the historical model product using effect information sequence to obtain a historical model product calling quantum sequence set and a historical model product using effect information subsequence set.
In some embodiments, the executing body may divide the historical model product call amount sequence and the historical model product use effect information sequence according to the trend difference information, to obtain a historical model product call quantum sequence set and a historical model product use effect information subsequence set. And the historical model product calling quantum sequences in the historical model product calling quantum sequence set and the historical model product using effect information subsequences in the historical model product using effect information subsequence set have a one-to-one correspondence.
As an example, the executing body may determine a historical model product calling quantum sequence and a historical model product using effect information subsequence corresponding to each trend difference sub-information in the plurality of trend difference sub-information, to obtain a historical model product calling quantum sequence set and a historical model product using effect information subsequence set. The time period corresponding to the trend difference sub-information, the time period corresponding to the historical model product calling quantum sequence and the time period corresponding to the historical model product using effect information sub-sequence are identical.
And 105, screening out a historical model product calling quantum sequence corresponding to the first trend type from the historical model product calling quantum sequence set to obtain at least one first historical model product calling quantum sequence.
In some embodiments, the executing body may screen the historical model product calling quantum sequence corresponding to the first trend type from the historical model product calling quantum sequence set to obtain at least one first historical model product calling quantum sequence. The first trend type may be preset, and represents a type that a call change trend corresponding to a call quantum sequence of the historical model product and an effect change trend corresponding to a using effect information subsequence of the historical model product are downward trend or upward trend. That is, the call amount corresponding to the historical model product call sub-sequence gradually decreases with time and the use effect corresponding to the historical model product use effect information sub-sequence gradually decreases with time, or the call amount corresponding to the historical model product call sub-sequence gradually increases with time and the use effect corresponding to the historical model product use effect information sub-sequence gradually increases with time. That is, the call volume corresponding to the first history model product call quantum sequence in the at least one first history model product call quantum sequence is gradually decreased with time or gradually increased with time.
And 106, determining at least one first historical model product use effect information subsequence corresponding to the at least one first historical model product calling quantum sequence in time.
In some embodiments, the executing entity may determine at least one first historical model product usage effect information subsequence that has a temporal correspondence with the at least one first historical model product invocation quantum sequence. The first historical model product use effect information subsequence in the at least one first historical model product use effect information subsequence has the same time period relation with the first historical model product call quantum sequence in the at least one first historical model product call quantum sequence. Similarly, the trend transformation between the first historical model product calling quantum sequence and the corresponding first historical model product using effect information subsequence is the same, namely the corresponding calling amount of the first historical model product calling quantum sequence gradually decreases with time and the corresponding first historical model product using effect information subsequence gradually decreases with time, or the corresponding calling amount of the first historical model product calling quantum sequence gradually increases with time and the corresponding first historical model product using effect information subsequence gradually increases with time.
And 107, generating data product model flow prediction information aiming at the target future time according to the at least one first historical model product calling quantum sequence, the at least one first historical model product using effect information sub-sequence and the model product information.
In some embodiments, the execution body may generate the data product model traffic prediction information for the target future time based on the at least one first historical model product invocation quantum sequence, the at least one first historical model product usage effect information sub-sequence, and the model product information.
As an example, the execution entity may generate the data product model traffic prediction information for the target future time according to the at least one first historical model product invocation quantum sequence, the at least one first historical model product usage effect information sub-sequence, and the model product information through various implementations.
In some optional implementations of some embodiments, generating the data product model traffic prediction information for the target future time based on the at least one first historical model product invocation quantum sequence, the at least one first historical model product usage effect information sub-sequence, and the model product information may include:
And a first step of inputting the at least one first historical model product call quantum sequence and the model product information into a first model flow prediction information generation model to generate first model flow prediction information for the target future time. Wherein the first model traffic prediction information generation model may be a neural network model that generates model traffic prediction information for a target future time. In practice, the first model traffic prediction information generation model may be a time-series neural network model. The first model flow prediction information may be predicted flow information of the target data model product for a target future time.
And a second step of inputting the at least one first historical model product usage effect information subsequence and the model product information into a second model flow prediction information generation model to generate second model flow prediction information for the target future time. Wherein the second model traffic prediction information generation model may be a neural network model that generates model traffic prediction information for the target future time. In practice, the second model traffic prediction information generation model may be a time-series neural network model. The second model flow prediction information may be predicted flow information for a target data model product for a target future time.
And thirdly, determining a first parameter value for the calling quantity of the first historical model product and a second parameter value for the using effect information of the first historical model product by using the first multi-head attention mechanism model. The first multi-headed attentiveness mechanism model may be a multi-headed attentiveness mechanism model in a transducer model.
And a fourth step of multiplying the first parameter value and the first model flow prediction information to obtain a first multiplication result, and multiplying the second parameter value and the second model flow prediction information to obtain a second multiplication result.
And fifthly, adding the first multiplication result and the second multiplication result to obtain an addition result which is used as the flow prediction information of the first candidate data product model.
In some optional implementations of some embodiments, the generating the data product model traffic prediction information for the target future time based on the at least one first historical model product invocation quantum sequence, the at least one first historical model product usage effect information sub-sequence, and the model product information further includes the steps of:
and screening out a historical model product calling quantum sequence corresponding to the second trend type from the historical model product calling quantum sequence set to obtain at least one second historical model product calling quantum sequence.
The second trend type may be preset, and is a type that characterizes a downward trend of a call change trend corresponding to a call quantum sequence of the historical model product and an upward trend of an effect change trend corresponding to a use effect information subsequence of the historical model product. That is, the call volume corresponding to the historical model product call sub-sequence gradually decreases with time and the use effect corresponding to the historical model product use effect information sub-sequence gradually increases with time. That is, the call volume corresponding to the second historical model product call quantum sequence in the at least one second historical model product call quantum sequence gradually decreases with time.
And a second step of determining at least one second historical model product use effect information subsequence corresponding to the at least one second historical model product calling quantum sequence in time.
That is, the second historical model product invocation quantum sequence of the at least one second historical model product invocation quantum sequence has the same time period relationship as the second historical model product usage effect information subsequence of the at least one second historical model product usage effect information subsequence.
And thirdly, inputting the at least one second historical model product calling quantum sequence and the model product information into a third model flow prediction information generation model to generate third model flow prediction information aiming at the target future time. Wherein the third model traffic prediction information generation model may be a neural network model that generates model traffic prediction information for the target future time. In practice, the third model traffic prediction information generation model may be a time-series neural network model. The third model flow prediction information may be predicted flow information for a target data model product for a target future time.
And a fourth step of inputting the at least one second historical model product usage effect information subsequence and the model product information into a fourth model flow prediction information generation model to generate fourth model flow prediction information for the target future time. Wherein the fourth model traffic prediction information generation model may be a neural network model that generates model traffic prediction information for the target future time. In practice, the fourth model traffic prediction information generation model may be a time-series neural network model. The fourth model flow prediction information may be predicted flow information for the target data model product for the target future time.
And fifthly, determining a third parameter value for the call quantity of the second historical model product and a fourth parameter value for the using effect information of the second historical model product by using the first multi-head attention mechanism model. Wherein the third parameter value may characterize a degree of importance of the second historical model product call volume. The fourth parameter value may characterize a degree of importance of the second historical model product usage effect information.
And a sixth step of multiplying the third parameter value by the third model flow prediction information to obtain a third multiplication result, and multiplying the fourth parameter value by the fourth model flow prediction information to obtain a fourth multiplication result.
And seventh, adding the third multiplication result and the fourth multiplication result to obtain an addition result, wherein the addition result is used as second candidate data product model flow prediction information.
In some optional implementations of some embodiments, the inputting the at least one second historical model product invocation quantum sequence and the model product information into a third model traffic prediction information generation model to generate third model traffic prediction information for the target future time may include the steps of:
and a first step of determining at least one historical time period corresponding to the at least one second historical model product calling quantum sequence.
And a second step of determining a difference historical time period between the at least one historical time period to obtain at least one difference historical time period. For example, the at least one second historical model product invocation quantum sequence includes: the second historical model product call quantum sequence A, the second historical model product call quantum sequence B, the second historical model product call quantum sequence C, the second historical model product call quantum sequence D and the second historical model product call quantum sequence E. The second historical model product calls the quantum sequence A to correspond to the time period of 1 month 1 day-1 month 7 days. The second historical model product calls the quantum sequence D to correspond to the time period of [1 month 10 days-1 month 12 days ]. The second historical model product calls the quantum sequence E to correspond to the time period of [1 month 17 day-1 month 21 day ]. Then, the at least one difference history period comprises: time period (7 days 1 month, 10 days 1 month), time period (12 days 1 month to 17 days 1 month).
And thirdly, sequencing the at least one difference historical time period to obtain a difference historical time period sequence.
As an example, the execution body may sort the at least one difference history period according to a time sequence, to obtain a difference history period sequence.
Fourth, for the difference history period in the series of difference history periods, the following generating step is performed:
the first substep, determining a third historical model product invocation quantum sequence preceding the difference historical time period. Wherein the third historical model product invocation quantum sequence includes at least one of: and the second historical model product calling quantum sequence set in the at least one second historical model product calling quantum sequence and the prediction generating product calling quantum sequence. The prediction generation product calling quantum sequence is generated based on a third model flow prediction information generation model. The time periods corresponding to the second historical model product calling quantum sequences in the second historical model product calling quantum sequence set in the at least one second historical model product calling quantum sequence are all before the difference historical time periods. The corresponding time period of the predicted generation product call quantum sequence is also before the difference time period.
For example, for a difference historical time period of time period (1 month 12 days-1 month 17 days), then the third historical model product invoking the quantum sequence includes: and the second historical model product calls the quantum sequence A, the prediction corresponding to the time period (1 month, 7 days, 1 month and 10 days) generates the product call quantum sequence, and the second historical model product calls the quantum sequence D. For the difference historical time period being a time period (1 month, 7 days, 1 month, 10 days), the third historical model product invoking quantum sequence comprises: the second historical model product invokes quantum sequence a. The prediction generation product call quantum sequence corresponding to the time period (1 month, 7 days, 1 month, and 10 days) may be generated by inputting the second historical model product call quantum sequence a into the third model traffic prediction information generation model.
And a second substep, carrying out data preprocessing on the call quantity of each historical model product in the third historical model product call quantum sequence to obtain a processed historical model product call quantum sequence.
As an example, the executing body may perform data supplementation and data normalization on each historical model product call amount in the third historical model product call quantum sequence to generate the processed historical model product call quantum sequence.
And a third substep, correspondingly multiplying each calling amount in the processed historical model product calling quantum sequence with each weight in a preset weight matrix list to obtain a first multiplication result sequence, wherein each weight in the weight matrix list is sequentially reduced, each weight in the weight matrix list shows an arithmetic progression relation, and the sum of each weight in the weight matrix list is a value of 1. For example, the weight matrix list may be [0.6,0.3,0.1,0.05,0.03,0.02]. The number of the weights included in the weight matrix list is the same as the number of the processed historical model product call quantities included in the processed historical model product call quantum sequence.
And a fourth substep, inputting the first multiplication result sequence into a product calling quantity prediction model included in a third model flow prediction information generation model to generate a prediction generation product calling quantum sequence aiming at a difference historical time period, and taking the product calling quantum sequence as a target prediction generation product calling quantum sequence. The product call quantity prediction model may be a neural network model for predicting the product call quantity in the target time period. Specifically, the product call volume prediction model may be an LSTM neural network model.
And a fifth substep, combining the product calling quantum sequence generated by target prediction with the third historical model product calling quantum sequence to obtain a combined model product calling quantum sequence.
And a sixth substep, in response to determining that the difference historical time period is the last time period, correspondingly multiplying each calling amount in the calling quantum sequence of the combined model product by each weight in the weight matrix list to obtain a second multiplication result sequence.
And a seventh sub-step of inputting a second multiplication result sequence into a flow prediction model included in the third model flow prediction information generation model to generate third model flow prediction information for the target future time. The flow prediction model may be a neural network model that predicts flow information at a future time. Specifically, the traffic prediction model may be an encoding and decoding neural network model. The coding model in the coding and decoding neural network model is a plurality of RNN models connected in series. The decoding model in the encoding and decoding neural network model may be a plurality of RNN models+a plurality of deconvolution neural network models+a fully connected layer connected in series.
And fifthly, responding to the fact that the time period is not the last time period, taking the next difference historical time period corresponding to the difference historical time period as the difference historical time period, taking the combined model product calling quantum sequence as a third historical model product calling quantum sequence corresponding to the difference historical time period, and continuing to execute the generating step.
Aiming at the technical problems that: the time period corresponding to the historical model product calling quantum sequence may not be continuous, so that the characteristic information reflected by the historical model product calling quantum sequence is also not continuous, and is directly input into a subsequent model flow prediction information generation model, so that the extracted characteristic of the model flow prediction information generation model is limited and discontinuous, and the flow prediction information at the future time is not accurate enough. In combination with the technical advantages/market state of knowledge possessed by the inventors and team, we decided to employ the following solutions:
in some optional implementations of some embodiments, the inputting the at least one second historical model product invocation quantum sequence and the model product information into a third model traffic prediction information generation model to generate third model traffic prediction information for the target future time may include the steps of:
and a first step of determining at least one historical time period corresponding to the at least one second historical model product calling quantum sequence.
And a second step of determining a difference historical time period between the at least one historical time period to obtain at least one difference historical time period.
And thirdly, sequencing the at least one difference historical time period to obtain a difference historical time period sequence.
Fourth, for the difference history period in the difference history period sequence, the following information generating step is performed:
and a first substep, in response to determining that the difference historical time period is not the time period corresponding to the last second historical model product calling quantum sequence in the at least one second historical model product calling quantum sequence, determining that a third historical model product calling quantum sequence corresponding to the historical time period before the difference historical time period and an adjacent historical model product calling quantum sequence corresponding to the historical time period after the difference historical time period and adjacent to the corresponding time period.
And a second sub-step of performing data preprocessing on the third historical model product calling quantum sequence and the adjacent historical model product calling quantum sequence to generate a processed historical model product calling quantum sequence and a processed adjacent historical model product calling quantum sequence.
And a third substep, correspondingly multiplying each calling amount in the processed historical model product calling quantum sequence with each weight in the first weight matrix list to obtain a fifth multiplication result sequence. Wherein, each weight in the first weight matrix list sequentially becomes larger, each weight in the first weight matrix list presents an arithmetic progression relation, and the sum of each weight in the first weight matrix list is a value of 1. The weights in the first weight matrix list may characterize the importance of the call volume in the call quantum sequence of the processed history model product relative to the call quantum sequence within the difference history time period.
And a fourth sub-step, correspondingly multiplying each calling amount in the processed adjacent historical model product calling quantum sequence with each weight in the second weight matrix list to obtain a geographic multiplication result sequence. Wherein, each weight in the second weight matrix list sequentially becomes smaller, each weight in the second weight matrix list presents an arithmetic progression relation, and the sum of each weight in the second weight matrix list is a value of 1. The number of the weights in the second weight matrix list is the same as the number of the historical model product calling quantities included in the processed adjacent historical model product calling quantum sequence. The weights in the second weight matrix list can represent the importance degree of the call quantity in the call quantum sequence of the processed adjacent history model product relative to the call quantum sequence in the difference history time period.
And a fifth substep, inputting the first multiplication result sequence into a product calling quantity prediction model included in a third model flow prediction information generation model to generate a prediction generation product calling quantum sequence aiming at a difference historical time period, and taking the product calling quantum sequence as a first target prediction generation product calling quantum sequence.
And a sixth substep, inputting the second multiplication result sequence into a historical product calling quantity prediction model included in a third model flow prediction information generation model to generate a historical prediction generation product calling quantum sequence aiming at a difference historical time period, and taking the historical prediction generation product calling quantum sequence as a target historical prediction generation product calling quantum sequence. The historical product call quantity prediction model may be a neural network model for predicting the product call quantity in the historical time. In practice, the historical product call volume prediction model may be a time-series neural network model.
And a seventh substep, inputting the first multiplication result sequence and the second multiplication result sequence into a target product calling quantity prediction model included in a third model flow prediction information generation model to generate a prediction generation product calling quantum sequence aiming at a difference historical time period, wherein the prediction generation product calling quantum sequence is used as a second target prediction generation product calling quantum sequence.
A seventh substep of generating a first model weight for the product call volume prediction model, a second model weight for the historical product call volume prediction model, and a third model weight for the target product call volume prediction model using a third multi-headed attention mechanism model.
And an eighth sub-step of multiplying the first target prediction generation product calling quantum sequence and the first model weight to obtain a fifth multiplication result sub-sequence, multiplying the target history prediction generation product calling quantum sequence and the second model weight to obtain a sixth multiplication result sub-sequence, and multiplying the second target prediction generation product calling quantum sequence and the third model weight to obtain a seventh multiplication result sub-sequence.
And a ninth sub-step of adding the corresponding positions of the multiplication results of the fifth multiplication result sub-sequence, the sixth multiplication result sub-sequence and the seventh multiplication result sub-sequence to obtain an addition result sub-sequence.
And a tenth sub-step of combining the added result sub-sequence with the third historical model product calling quantum sequence to obtain a target combined model product calling quantum sequence.
And an eleventh substep, in response to determining that the difference historical time period is the last time period, correspondingly multiplying each calling amount in the target combination model product calling quantum sequence by each weight in the third weight matrix list to obtain an eighth multiplication result sequence. Wherein, each weight in the third weight matrix list sequentially becomes larger, each weight in the third weight matrix list presents an arithmetic progression relation, and the sum of each weight in the third weight matrix list is a value of 1. Wherein the number of the respective weights in the third weight matrix list is the same as the number of the eighth multiplication results included in the eighth multiplication result sequence. The weights in the third weight matrix list characterize the importance of the multiplication results in the eighth multiplication result sequence with respect to the third model flow prediction information for the target future time.
A twelfth substep of inputting an eighth multiplication result sequence into a flow prediction model included in the third model flow prediction information generation model to generate third model flow prediction information for the target future time.
And fifthly, responding to the fact that the time period is not the last time period, taking the next difference historical time period corresponding to the difference historical time period as the difference historical time period, taking the target combination model product calling quantum sequence as a third history model product calling quantum sequence corresponding to the difference historical time period, and continuing to execute the information generation step.
In some optional implementations of some embodiments and optional matters, as an invention point of the disclosure, the technical problem mentioned in the background art is solved, that "a time period corresponding to a historical model product calling quantum sequence may not be continuous, so that feature information reflected by the historical model product calling quantum sequence is also not continuous, and is directly input into a subsequent model flow prediction information generation model, so that features extracted by the model flow prediction information generation model are limited and discontinuous, and the obtained flow prediction information at a future time is not accurate enough. Based on the above, firstly, the method and the device make up for the characteristic information of the missing part of time periods in the historical model product calling quantum sequence by predicting the model product calling amount corresponding to at least one difference historical time period, so that the subsequent third model flow prediction information generation model can learn more characteristic information, and flow prediction information in future time can be accurately generated. In addition, by setting the first multiplication result sequence and the second multiplication result sequence, the flow prediction information at a more accurate future time can be obtained later in consideration of the influence of time on the subtraction of the characteristic information.
In some optional implementations of some embodiments, the inputting the at least one second historical model product usage effect information subsequence and the model product information into a fourth model flow prediction information generation model to generate fourth model flow prediction information for the target future time may include the steps of:
and a first step of determining at least one historical time period corresponding to the at least one second historical model product using effect information subsequence.
And a second step of determining a difference historical time period between the at least one historical time period to obtain at least one difference historical time period. For example, the at least one second historical model product usage effect information subsequence includes: the second historical model product usage effect information sub-sequence A, the second historical model product usage effect information sub-sequence D, and the second historical model product usage effect information sub-sequence E. The time period corresponding to the second historical model product using effect information subsequence A is [1 month 1 day-1 month 7 days ]. The second historical model product usage effect information subsequence D corresponds to a time period of [1 month 10 days-1 month 12 days ]. The second historical model product usage effect information subsequence E corresponds to a time period of [1 month 17 days-1 month 21 days ]. Then, the at least one difference history period comprises: time period (7 days 1 month, 10 days 1 month), time period (12 days 1 month to 17 days 1 month).
And thirdly, sequencing the at least one difference historical time period to obtain a difference historical time period sequence.
As an example, the execution body may sort the at least one difference history period according to a time sequence, to obtain a difference history period sequence.
Fourth, for the difference history period in the series of difference history periods, the following generating step is performed:
a first sub-step of determining a third historical model product usage effect information sub-sequence preceding the difference historical time period. Wherein the third historical model product usage effect information subsequence includes at least one of: and a second historical model product use effect information subsequence set in the at least one second historical model product use effect information subsequence and a prediction generation model product use effect information subsequence. The prediction generation model product use effect information subsequence is generated based on the fourth model flow prediction information generation model. The time periods corresponding to the second historical model product usage effect information subsequences in the second historical model product usage effect information subsequence set are all before the difference historical time period. The corresponding time period of the predicted generation model product using effect information subsequence is also before the difference time period.
For example, for a difference historical time period of time (1 month 12 days-1 month 17 days), then the third historical model product usage effect information subsequence includes: the second historical model product use effect information subsequence A, the prediction corresponding to the time period (1 month, 7 days, 1 month and 10 days) generate model product use effect information subsequence and the second historical model product use effect information subsequence D. For the difference historical time period being a time period (1 month, 7 days, 1 month, 10 days), the third historical model product usage effect information subsequence includes: the second historical model product uses the effect information subsequence a. The prediction generation model product use effect information subsequence corresponding to the time period (1 month, 7 days, 1 month, and 10 days) may be generated by inputting the second historical model product use effect information subsequence a into the fourth model flow prediction information generation model.
And a second sub-step, performing data preprocessing on the use effect information of each historical model product in the third historical model product use effect information sub-sequence to obtain a processed historical model product use effect information sub-sequence.
As an example, the executing body may perform data supplementation and data normalization on each of the history model product usage effect information in the third history model product usage effect information subsequence to generate the processed history model product usage effect information subsequence.
And a third sub-step of correspondingly multiplying each history model product use effect information in the processed history model product use effect information subsequence with each weight in a preset weight matrix list to obtain a third multiplication result sequence, wherein each weight in the preset weight matrix list is sequentially reduced, each weight in the preset weight matrix list shows an arithmetic series relationship, and the sum of each weight in the preset weight matrix list is a value of 1. For example, the predetermined weight matrix list may be [0.5,0.4,0.1,0.05,0.03,0.02]. The number of the weights included in the predetermined weight matrix list is the same as the number of the post-processing historical model product call quantities included in the post-processing historical model product call quantum sequence.
And a fourth sub-step of inputting the third multiplication result sequence into a model product use effect information prediction model included in a fourth model flow prediction information generation model to generate a prediction generation model product use effect information sub-sequence aiming at the difference historical time period, and taking the prediction generation model product use effect information sub-sequence as a target prediction generation model product use effect information sub-sequence. The model product use effect information prediction model may be a neural network model for predicting model product use effect information in a target time period. In particular, the model product usage effect information prediction model may be an LSTM neural network model.
And a fifth sub-step of combining the target prediction generation model product use effect information sub-sequence with the third historical model product use effect information sub-sequence to obtain a combined model product use effect information sub-sequence.
And a sixth sub-step, in response to determining that the difference historical time period is the last time period, correspondingly multiplying each calling amount in the combined model product using effect information subsequence by each weight in the preset weight matrix list to obtain a fourth multiplication result sequence.
And a seventh sub-step of inputting a fourth multiplication result sequence into a flow prediction model included in the fourth model flow prediction information generation model to generate fourth model flow prediction information for the target future time. The flow prediction model may be a neural network model that predicts flow information at a future time. Specifically, the traffic prediction model may be an encoding and decoding neural network model. The coding model in the coding and decoding neural network model is a plurality of RNN models connected in series. The decoding model in the encoding and decoding neural network model may be a plurality of RNN models+a plurality of deconvolution neural network models+a fully connected layer connected in series.
And fifthly, in response to determining that the time period is not the last time period, taking the next difference historical time period corresponding to the difference historical time period as the difference historical time period, taking the combined model product use effect information subsequence as a third historical model product use effect information subsequence corresponding to the difference historical time period, and continuing to execute the generating step.
In some optional implementations of some embodiments, the generating the data product model traffic prediction information for the target future time based on the at least one first historical model product invocation quantum sequence, the at least one first historical model product usage effect information sub-sequence, and the model product information further includes the steps of:
screening historical model product calling quantum sequences corresponding to the third trend type from the historical model product calling quantum sequence set to obtain at least one third historical model product calling quantum sequence. The third trend type may be a preset type, which characterizes that a call change trend corresponding to a call quantum sequence of the historical model product is an upward trend, and an effect change trend corresponding to a use effect information subsequence of the historical model product is a downward trend. That is, the call volume corresponding to the historical model product call sub-sequence gradually increases with time and the use effect corresponding to the historical model product use effect information sub-sequence gradually decreases with time. That is, the call volume corresponding to the third history model product call quantum sequence in the at least one third history model product call quantum sequence is gradually increased with time.
And a second step of determining at least one third historical model product use effect information subsequence corresponding to the at least one third historical model product calling quantum sequence in time. The third historical model product calling quantum sequence in the at least one third historical model product calling quantum sequence has the same time period relationship with the third historical model product using effect information subsequence in the at least one third historical model product using effect information subsequence.
And thirdly, inputting the at least one third historical model product calling quantum sequence and the model product information into a fifth model flow prediction information generation model to generate fifth model flow prediction information aiming at the target future time. Wherein the fifth model traffic prediction information generation model may be a neural network model that generates model traffic prediction information for the target future time. In practice, the fifth model traffic prediction information generation model may be a time-series neural network model. The fifth model flow prediction information may be predicted flow information of the target data model product for the target future time.
And a fourth step of inputting the at least one third historical model product usage effect information subsequence and the model product information into a sixth model flow prediction information generation model to generate sixth model flow prediction information for the target future time. Wherein the sixth model traffic prediction information generation model may be a neural network model that generates model traffic prediction information for the target future time. In practice, the sixth model traffic prediction information generation model may be a time-series neural network model. The sixth model flow prediction information may be predicted flow information for the target data model product for the target future time.
And fifthly, determining a fifth parameter value for the calling quantity of the third historical model product and a sixth parameter value for the using effect information of the third historical model product by using the first multi-head attention mechanism model. Wherein the fifth parameter value may characterize a degree of importance of the third historical model product call volume. The sixth parameter value may characterize a degree of importance of the third historical model product usage effect information.
And a sixth step of multiplying the fifth parameter value by the fifth model flow prediction information to obtain a fifth multiplication result, and multiplying the sixth parameter value by the sixth model flow prediction information to obtain a sixth multiplication result.
And seventh, adding the fifth multiplication result and the sixth multiplication result to obtain an addition result, wherein the addition result is used as third candidate data product model flow prediction information.
And eighth step, determining a first type weight corresponding to the first trend type, a second type weight corresponding to the second trend type and a third type weight corresponding to the third trend type by using a second multi-head attention mechanism model. The second multi-head attention mechanism model may be a multi-head attention mechanism model in a transform model, and is configured to output a first type weight corresponding to the first trend type, a second type weight corresponding to the second trend type, and a third type weight corresponding to the third trend type.
And a ninth step of multiplying the first candidate data product model traffic prediction information by the first type weight to generate a seventh multiplication result, multiplying the second candidate data product model traffic prediction information by the second type weight to generate an eighth multiplication result, and multiplying the third candidate data product model traffic prediction information by the third type weight to generate a ninth multiplication result.
And a tenth step of adding the seventh multiplication result, the eighth multiplication result, and the ninth multiplication result to generate an addition result as data product model flow prediction information for the target future time.
In some optional implementations of some embodiments, the first trend type characterizes a downward trend or an upward trend of a call-change trend corresponding to the historical model product call quantum sequence and an effect-change trend corresponding to the historical model product use effect information subsequence, the second trend type characterizes a downward trend of a call-change trend corresponding to the historical model product call quantum sequence and an upward trend of an effect-change trend corresponding to the historical model product use effect information subsequence, and the third trend type characterizes a upward trend of a call-change trend corresponding to the historical model product call quantum sequence and an effect-change trend corresponding to the historical model product use effect information subsequence.
And step 108, according to the data product model flow prediction information, value delivery information of corresponding volumes is delivered to a target client so as to display the value delivery information on the target client.
In some embodiments, the executing entity may put the value put information corresponding to the volume at the target client according to the data product model flow prediction information, so as to display the value put information at the target client. In practice, the value impression information may be advertisements for the targeted data model product for the electronic marketplace scenario.
As an example, the executing entity may obtain the volume of the value delivery information corresponding to the data product model flow prediction information. And then, putting the value putting information of the corresponding volume in the target client so as to show the value putting information in the target client.
The above embodiments of the present disclosure have the following advantageous effects: the product model flow information is quickly and accurately generated by the product model flow information generation method of some embodiments of the present disclosure, so that the subsequent accurate delivery of value to the target data product model is facilitated. Specifically, the reason why the accurate delivery of value to the target data product model is not possible is: the product flow is predicted manually only through various product indexes intuitively, and the product model flow information generated is not accurate enough due to the fact that the product flow is large in one-sided performance, so that accurate value throwing cannot be performed on target data model products in the follow-up process. Based on this, in the product model flow information generating method of some embodiments of the present disclosure, first, a historical model product call amount sequence, a historical model product use effect information sequence and model product information corresponding to a target data model product are obtained for prediction of data product model flow prediction information at a subsequent target future time. And then determining call amount transformation trend information corresponding to the call amount sequence of the historical model product and determining effect transformation trend information corresponding to the using effect information sequence of the historical model product. Here, the calling amount change trend information and the effect change trend information are determined to determine the difference between the calling amount change trend information and the effect change trend information, so that prediction supplementation of corresponding trend information is performed according to the difference between the calling amount change trend information and the effect change trend information, and the situation that larger errors occur in the process of generating data product model flow prediction information in the follow-up process due to information errors in the history model calling amount sequence and the history model product use effect information sequence is avoided. And then, determining trend difference information between the call volume transformation trend information and the effect transformation trend information as an actual information sequence for predicting the model product call volume and the model product utilization effect information of the subsequent rest time period. Wherein, the difference time length corresponding to the trend difference information is longer than the target time length. And then, according to the trend difference information, carrying out sequence division on the historical model product calling quantity sequence and the historical model product using effect information sequence to obtain a historical model product calling quantum sequence set and a historical model product using effect information subsequence set, wherein the historical model product calling quantum sequence set and the historical model product using effect information subsequence set have a one-to-one correspondence. Here, through the sequence division, the call volume sequence and the usage effect information subsequence corresponding to the trend type are determined later. And screening out a historical model product calling quantum sequence corresponding to the first trend type from the historical model product calling quantum sequence set to obtain at least one first historical model product calling quantum sequence for generating flow prediction information of a subsequent data product model. And further determining at least one first historical model product use effect information subsequence in a time corresponding relation with the at least one first historical model product calling quantum sequence, so as to be used for generating flow prediction information of a subsequent data product model. Further, based on the at least one first historical model product invocation quantum sequence, the at least one first historical model product usage effect information sub-sequence, and the model product information, data product model flow prediction information for a target future time may be accurately generated. And finally, according to the data product model flow prediction information, value delivery information corresponding to the volume is delivered to a target client side, so that the value delivery information is displayed on the target client side, and the model related content of the data product model is displayed more widely.
With further reference to fig. 2, as an implementation of the method shown in the above figures, the present disclosure provides some embodiments of a product model flow information generating apparatus, which correspond to those method embodiments shown in fig. 1, and which are particularly applicable to various electronic devices.
As shown in fig. 2, a product model flow information generating apparatus 200 includes: an acquisition unit 201, a first determination unit 202, a second determination unit 203, a division unit 204, a screening unit 205, a third determination unit 206, a generation unit 207, and a delivery unit 208. Wherein, the obtaining unit 201 is configured to obtain a historical model product calling amount sequence, a historical model product using effect information sequence and model product information corresponding to the target data model product; a first determining unit 202 configured to determine call amount transformation trend information corresponding to the call amount sequence of the history model product, and determine effect transformation trend information corresponding to the effect information sequence of the history model product; a second determining unit 203 configured to determine trend difference information between the call-up conversion trend information and the effect conversion trend information, wherein a difference time period corresponding to the trend difference information is longer than a target time period; the dividing unit 204 is configured to divide the historical model product calling amount sequence and the historical model product using effect information sequence according to the trend difference information to obtain a historical model product calling quantum sequence set and a historical model product using effect information subsequence set, wherein the historical model product calling quantum sequence set and the historical model product using effect information subsequence set have a one-to-one correspondence in time; the screening unit 205 is configured to screen out a historical model product calling quantum sequence corresponding to the first trend type from the historical model product calling quantum sequence set, so as to obtain at least one first historical model product calling quantum sequence; a third determining unit 206 configured to determine at least one first historical model product usage effect information subsequence having a temporal correspondence with the at least one first historical model product invocation quantum sequence; a generating unit 207 configured to generate data product model flow prediction information for a target future time based on the at least one first historical model product invocation quantum sequence, the at least one first historical model product usage effect information sub-sequence, and the model product information; and the delivering unit 208 is configured to deliver the value delivering information corresponding to the volume at the target client according to the data product model flow prediction information, so as to display the value delivering information at the target client.
It will be appreciated that the elements described in the product model flow information generation apparatus 200 correspond to the various steps in the method described with reference to fig. 1. Thus, the operations, features and advantages described above with respect to the method are equally applicable to the product model flow information generating apparatus 200 and the units contained therein, and are not described herein.
Referring now to fig. 3, a schematic diagram of an electronic device (e.g., electronic device) 300 suitable for use in implementing some embodiments of the present disclosure is shown. The electronic device shown in fig. 3 is merely an example and should not impose any limitations on the functionality and scope of use of embodiments of the present disclosure.
As shown in fig. 3, the electronic device 300 may include a processing means (e.g., a central processing unit, a graphics processor, etc.) 301 that may perform various suitable actions and processes in accordance with a program stored in a Read Only Memory (ROM) 302 or a program loaded from a storage means 308 into a Random Access Memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the electronic apparatus 300 are also stored. The processing device 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input/output (I/O) interface 305 is also connected to bus 304.
In general, the following devices may be connected to the I/O interface 305: input devices 306 including, for example, a touch screen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; an output device 307 including, for example, a Liquid Crystal Display (LCD), a speaker, a vibrator, and the like; storage 308 including, for example, magnetic tape, hard disk, etc.; and communication means 309. The communication means 309 may allow the electronic device 300 to communicate with other devices wirelessly or by wire to exchange data. While fig. 3 shows an electronic device 300 having various means, it is to be understood that not all of the illustrated means are required to be implemented or provided. More or fewer devices may be implemented or provided instead. Each block shown in fig. 3 may represent one device or a plurality of devices as needed.
In particular, according to some embodiments of the present disclosure, the processes described above with reference to flowcharts may be implemented as computer software programs. For example, some embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method shown in the flow chart. In such embodiments, the computer program may be downloaded and installed from a network via communications device 309, or from storage device 308, or from ROM 302. The above-described functions defined in the methods of some embodiments of the present disclosure are performed when the computer program is executed by the processing means 301.
It should be noted that, in some embodiments of the present disclosure, the computer readable medium may be a computer readable signal medium or a computer readable storage medium, or any combination of the two. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In some embodiments of the present disclosure, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In some embodiments of the present disclosure, however, the computer-readable signal medium may comprise a data signal propagated in baseband or as part of a carrier wave, with the computer-readable program code embodied therein. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: electrical wires, fiber optic cables, RF (radio frequency), and the like, or any suitable combination of the foregoing.
In some implementations, the clients, servers may communicate using any currently known or future developed network protocol, such as HTTP (HyperText Transfer Protocol ), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), the internet (e.g., the internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future developed networks.
The computer readable medium may be contained in the electronic device; or may exist alone without being incorporated into the electronic device. The computer readable medium carries one or more programs which, when executed by the electronic device, cause the electronic device to: acquiring a historical model product calling amount sequence corresponding to a target data model product, a historical model product using effect information sequence and model product information; determining call amount change trend information corresponding to the call amount sequence of the historical model product and effect change trend information corresponding to the use effect information sequence of the historical model product; determining trend difference information between the call level change trend information and the effect change trend information, wherein the difference time length corresponding to the trend difference information is longer than a target time length; according to the trend difference information, carrying out sequence division on the historical model product calling quantity sequence and the historical model product using effect information sequence to obtain a historical model product calling quantum sequence set and a historical model product using effect information subsequence set, wherein the historical model product calling quantum sequence set and the historical model product using effect information subsequence set have a one-to-one correspondence; screening historical model product calling quantum sequences corresponding to the first trend types from the historical model product calling quantum sequence set to obtain at least one first historical model product calling quantum sequence; determining at least one first historical model product use effect information subsequence corresponding to the at least one first historical model product calling quantum sequence in time; generating data product model flow prediction information for a target future time according to the at least one first historical model product call quantum sequence, the at least one first historical model product use effect information subsequence and the model product information; and according to the data product model flow prediction information, putting value putting information corresponding to the volume in a target client so as to show the value putting information in the target client.
Computer program code for carrying out operations for some embodiments of the present disclosure may be written in one or more programming languages, including an object oriented programming language such as Java, smalltalk, C ++ and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computer (for example, through the Internet using an Internet service provider).
The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The units described in some embodiments of the present disclosure may be implemented by means of software, or may be implemented by means of hardware. The described units may also be provided in a processor, for example, described as: the processor comprises an acquisition unit, a first determination unit, a second determination unit, a division unit, a screening unit, a third determination unit, a generation unit and a delivery unit. The names of these units do not constitute limitations on the unit itself in some cases, and for example, the acquisition unit may also be described as "a unit that acquires a history model product call amount sequence, a history model product use effect information sequence, and model product information corresponding to a target data model product".
The functions described above herein may be performed, at least in part, by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: a Field Programmable Gate Array (FPGA), an Application Specific Integrated Circuit (ASIC), an Application Specific Standard Product (ASSP), a system on a chip (SOC), a Complex Programmable Logic Device (CPLD), and the like.
The foregoing description is only of the preferred embodiments of the present disclosure and description of the principles of the technology being employed. It will be appreciated by those skilled in the art that the scope of the invention in the embodiments of the present disclosure is not limited to the specific combination of the above technical features, but encompasses other technical features formed by any combination of the above technical features or their equivalents without departing from the spirit of the invention. Such as the above-described features, are mutually substituted with (but not limited to) the features having similar functions disclosed in the embodiments of the present disclosure.
Claims (9)
1. A method for generating flow information of a product model comprises the following steps:
acquiring a historical model product calling amount sequence corresponding to a target data model product, a historical model product using effect information sequence and model product information;
determining call amount conversion trend information corresponding to the call amount sequence of the historical model product and effect conversion trend information corresponding to the use effect information sequence of the historical model product;
trend difference information between the call level change trend information and the effect change trend information is determined, wherein the difference time length corresponding to the trend difference information is longer than a target time length;
according to the trend difference information, carrying out sequence division on the historical model product calling quantity sequence and the historical model product using effect information sequence to obtain a historical model product calling quantum sequence set and a historical model product using effect information subsequence set, wherein the historical model product calling quantum sequence set and the historical model product using effect information subsequence set have a one-to-one correspondence;
screening historical model product calling quantum sequences corresponding to the first trend type from the historical model product calling quantum sequence set to obtain at least one first historical model product calling quantum sequence;
Determining at least one first historical model product use effect information subsequence in a time corresponding relationship with the at least one first historical model product call quantum sequence;
generating data product model flow prediction information for a target future time according to the at least one first historical model product call quantum sequence, the at least one first historical model product use effect information subsequence and the model product information;
and according to the data product model flow prediction information, value delivery information of corresponding volumes is delivered to a target client so as to display the value delivery information on the target client.
2. The method of claim 1, wherein the generating data product model traffic prediction information for a target future time from the at least one first historical model product invocation quantum sequence, the at least one first historical model product usage effect information subsequence, and the model product information comprises:
inputting the at least one first historical model product invocation quantum sequence and the model product information into a first model traffic prediction information generation model to generate first model traffic prediction information for the target future time;
Inputting the at least one first historical model product usage effect information subsequence and the model product information into a second model flow prediction information generation model to generate second model flow prediction information for the target future time;
determining a first parameter value for the call volume of the first historical model product and a second parameter value for the use effect information of the first historical model product by using the first multi-head attention mechanism model;
multiplying the first parameter value and the first model flow prediction information to obtain a first multiplication result, and multiplying the second parameter value and the second model flow prediction information to obtain a second multiplication result;
and adding the first multiplication result and the second multiplication result to obtain an addition result which is used as the flow prediction information of the first candidate data product model.
3. The method of claim 2, wherein the generating data product model traffic prediction information for a target future time from the at least one first historical model product invocation quantum sequence, the at least one first historical model product usage effect information subsequence, and the model product information, further comprises:
Screening historical model product calling quantum sequences corresponding to the second trend type from the historical model product calling quantum sequence set to obtain at least one second historical model product calling quantum sequence;
determining at least one second historical model product use effect information subsequence in a time corresponding relationship with the at least one second historical model product call quantum sequence;
inputting the at least one second historical model product invocation quantum sequence and the model product information into a third model traffic prediction information generation model to generate third model traffic prediction information for the target future time;
inputting the at least one second historical model product usage effect information subsequence and the model product information into a fourth model flow prediction information generation model to generate fourth model flow prediction information for the target future time;
determining a third parameter value for the call volume of the second historical model product and a fourth parameter value for the use effect information of the second historical model product by using the first multi-head attention mechanism model;
multiplying the third parameter value and the third model flow prediction information to obtain a third multiplication result, and multiplying the fourth parameter value and the fourth model flow prediction information to obtain a fourth multiplication result;
And adding the third multiplication result and the fourth multiplication result to obtain an addition result, and taking the addition result as second candidate data product model flow prediction information.
4. The method of claim 3, wherein the generating data product model traffic prediction information for a target future time from the at least one first historical model product invocation quantum sequence, the at least one first historical model product usage effect information subsequence, and the model product information, further comprises:
screening historical model product calling quantum sequences corresponding to a third trend type from the historical model product calling quantum sequence set to obtain at least one third historical model product calling quantum sequence;
determining at least one third historical model product use effect information subsequence in a time corresponding relationship with the at least one third historical model product call quantum sequence;
inputting the at least one third historical model product invocation quantum sequence and the model product information into a fifth model flow prediction information generation model to generate fifth model flow prediction information for the target future time;
Inputting the at least one third historical model product usage effect information subsequence and the model product information into a sixth model flow prediction information generation model to generate sixth model flow prediction information for the target future time;
determining a fifth parameter value for the call quantity of the third historical model product and a sixth parameter value for the use effect information of the third historical model product by using the first multi-head attention mechanism model;
multiplying the fifth parameter value and the fifth model flow prediction information to obtain a fifth multiplication result, and multiplying the sixth parameter value and the sixth model flow prediction information to obtain a sixth multiplication result;
adding the fifth multiplication result and the sixth multiplication result to obtain an addition result, and taking the addition result as third candidate data product model flow prediction information;
determining a first type weight corresponding to the first trend type, a second type weight corresponding to the second trend type and a third type weight corresponding to the third trend type by using a second multi-head attention mechanism model;
multiplying the first candidate data product model traffic prediction information with the first type of weight to generate a seventh multiplication result, multiplying the second candidate data product model traffic prediction information with the second type of weight to generate an eighth multiplication result, and multiplying the third candidate data product model traffic prediction information with the third type of weight to generate a ninth multiplication result;
And adding the seventh multiplication result, the eighth multiplication result and the ninth multiplication result to generate an addition result as data product model flow prediction information for the target future time.
5. The method of claim 4, wherein the first trend type characterizes a downward trend or an upward trend of a call-up change trend corresponding to the historical model product call quantum sequence and an effect change trend corresponding to the historical model product use effect information subsequence, the second trend type characterizes a downward trend of a call-up change trend corresponding to the historical model product call quantum sequence and an upward trend of an effect change trend corresponding to the historical model product use effect information subsequence, and the third trend type characterizes a upward trend of a call-up change trend corresponding to the historical model product call quantum sequence and an effect change trend corresponding to the historical model product use effect information subsequence.
6. The method of claim 5, wherein the inputting the at least one second historical model product invocation quantum sequence and the model product information into a third model traffic prediction information generation model to generate third model traffic prediction information for the target future time comprises:
Determining at least one historical time period corresponding to the at least one second historical model product call quantum sequence;
determining a difference historical time period between at least one historical time period to obtain at least one difference historical time period;
sorting the at least one difference historical time period to obtain a difference historical time period sequence;
for the difference history time period in the series of difference history time periods, the following generating step is performed:
determining a third historical model product invocation quantum sequence prior to the difference historical time period, wherein the third historical model product invocation quantum sequence includes at least one of: the second historical model product calling quantum sequence and the prediction generation product calling quantum sequence in the at least one second historical model product calling quantum sequence are generated based on a third model flow prediction information generation model;
performing data preprocessing on the call quantity of each historical model product in the third historical model product call quantum sequence to obtain a processed historical model product call quantum sequence;
correspondingly multiplying each calling amount in the processed historical model product calling quantum sequence with each weight in a preset weight matrix list to obtain a first multiplication result sequence, wherein each weight in the weight matrix list sequentially becomes smaller, each weight in the weight matrix list shows an arithmetic progression relation, and the sum of each weight in the weight matrix list is a value of 1;
Inputting the first multiplication result sequence into a product calling quantity prediction model included in a third model flow prediction information generation model to generate a prediction generation product calling quantum sequence aiming at a difference historical time period, and taking the prediction generation product calling quantum sequence as a target prediction generation product calling quantum sequence;
combining the product calling quantum sequence generated by target prediction with the third historical model product calling quantum sequence to obtain a combined model product calling quantum sequence;
responding to the fact that the historical time period of the difference value is the last time period, and correspondingly multiplying each calling amount in the calling quantum sequence of the combined model product by each weight in the weight matrix list to obtain a second multiplication result sequence;
inputting a second multiplication result sequence to a flow prediction model included in the third model flow prediction information generation model to generate third model flow prediction information for the target future time;
and in response to determining that the time period is not the last time period, taking the next difference historical time period corresponding to the difference historical time period as the difference historical time period, taking the combined model product calling quantum sequence as a third historical model product calling quantum sequence corresponding to the difference historical time period, and continuing to execute the generating step.
7. A product model flow information generation apparatus comprising:
the acquisition unit is configured to acquire a historical model product calling amount sequence, a historical model product using effect information sequence and model product information corresponding to the target data model product;
a first determining unit configured to determine call amount transformation trend information corresponding to the call amount sequence of the historical model product and determine effect transformation trend information corresponding to the effect information sequence of the historical model product;
a second determining unit configured to determine trend difference information between the call-up transformation trend information and the effect transformation trend information, wherein a difference time period corresponding to the trend difference information is longer than a target time period;
the dividing unit is configured to divide the historical model product calling quantity sequence and the historical model product using effect information sequence in sequence according to the trend difference information to obtain a historical model product calling quantum sequence set and a historical model product using effect information subsequence set, wherein the historical model product calling quantum sequence set and the historical model product using effect information subsequence set have a one-to-one correspondence in time;
The screening unit is configured to screen a historical model product calling quantum sequence corresponding to the first trend type from the historical model product calling quantum sequence set to obtain at least one first historical model product calling quantum sequence;
a third determining unit configured to determine at least one first historical model product usage effect information subsequence having a temporal correspondence with the at least one first historical model product invocation quantum sequence;
a generation unit configured to generate data product model flow prediction information for a target future time from the at least one first historical model product invocation quantum sequence, the at least one first historical model product usage effect information sub-sequence, and the model product information;
and the delivery unit is configured to deliver the value delivery information of the corresponding volume at the target client according to the data product model flow prediction information so as to display the value delivery information at the target client.
8. An electronic device, comprising:
one or more processors;
a storage device having one or more programs stored thereon,
when executed by the one or more processors, causes the one or more processors to implement the method of any of claims 1-6.
9. A computer readable medium having stored thereon a computer program, wherein the program when executed by a processor implements the method of any of claims 1-6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311674499.XA CN117689253A (en) | 2023-12-07 | 2023-12-07 | Method, device, equipment and medium for generating flow information of product model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311674499.XA CN117689253A (en) | 2023-12-07 | 2023-12-07 | Method, device, equipment and medium for generating flow information of product model |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117689253A true CN117689253A (en) | 2024-03-12 |
Family
ID=90131192
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311674499.XA Pending CN117689253A (en) | 2023-12-07 | 2023-12-07 | Method, device, equipment and medium for generating flow information of product model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117689253A (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120158456A1 (en) * | 2010-12-20 | 2012-06-21 | Xuerui Wang | Forecasting Ad Traffic Based on Business Metrics in Performance-based Display Advertising |
CN110688735A (en) * | 2019-09-05 | 2020-01-14 | 浪潮(北京)电子信息产业有限公司 | Time sequence signal trend prediction method, device, equipment and storage medium |
CN111130932A (en) * | 2019-12-18 | 2020-05-08 | 北京浩瀚深度信息技术股份有限公司 | Method and device for predicting flow trend based on historical flow and storage medium |
WO2023124469A1 (en) * | 2021-12-28 | 2023-07-06 | 中兴通讯股份有限公司 | Device energy saving method and system, electronic device, and storage medium |
-
2023
- 2023-12-07 CN CN202311674499.XA patent/CN117689253A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120158456A1 (en) * | 2010-12-20 | 2012-06-21 | Xuerui Wang | Forecasting Ad Traffic Based on Business Metrics in Performance-based Display Advertising |
CN110688735A (en) * | 2019-09-05 | 2020-01-14 | 浪潮(北京)电子信息产业有限公司 | Time sequence signal trend prediction method, device, equipment and storage medium |
CN111130932A (en) * | 2019-12-18 | 2020-05-08 | 北京浩瀚深度信息技术股份有限公司 | Method and device for predicting flow trend based on historical flow and storage medium |
WO2023124469A1 (en) * | 2021-12-28 | 2023-07-06 | 中兴通讯股份有限公司 | Device energy saving method and system, electronic device, and storage medium |
Non-Patent Citations (1)
Title |
---|
潘晨宇: "移动网络流量和5G用户数预测", 互联网技术, no. 2, 15 February 2022 (2022-02-15) * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN115293657B (en) | Carbon emission index information generation method, apparatus, electronic device, and medium | |
CN117236805B (en) | Power equipment control method, device, electronic equipment and computer readable medium | |
CN111932348B (en) | Alarm method and device for abnormal order, electronic equipment and readable medium | |
CN117035842A (en) | Model training method, traffic prediction method, device, equipment and medium | |
CN112182374B (en) | Inventory control method, apparatus, electronic device, and computer-readable medium | |
CN112989203B (en) | Material throwing method, device, equipment and medium | |
CN116388112A (en) | Abnormal supply end power-off method, device, electronic equipment and computer readable medium | |
CN117689253A (en) | Method, device, equipment and medium for generating flow information of product model | |
CN114157578B (en) | Network state prediction method and device | |
CN112019406B (en) | Flow monitoring method and device, electronic equipment and computer readable medium | |
CN114723455A (en) | Service processing method and device, electronic equipment and storage medium | |
CN111130984A (en) | Method and apparatus for processing information | |
CN114816955A (en) | Database performance prediction method and device | |
CN116416018A (en) | Content output method, content output device, computer readable medium and electronic equipment | |
CN114764627A (en) | Data contribution capacity determination method and device based on transverse joint learning participants | |
CN114792258B (en) | Information generation method and device, electronic equipment and computer readable medium | |
CN112073202B (en) | Information generation method and device, electronic equipment and computer readable medium | |
CN111949938B (en) | Determination method and device of transaction information, electronic equipment and computer readable medium | |
CN117236653A (en) | Traffic prediction-based vehicle scheduling method and device and electronic equipment | |
CN116800834B (en) | Virtual gift merging method, device, electronic equipment and computer readable medium | |
CN118052580A (en) | Model generation method, order quantity generation device, equipment and medium | |
CN115565607B (en) | Method, device, readable medium and electronic equipment for determining protein information | |
CN108959014B (en) | Method and apparatus for monitoring a platform | |
CN117876091A (en) | Information transmission method, apparatus, electronic device, and computer-readable medium | |
CN118052509A (en) | Inventory information generation method, apparatus, electronic device and computer readable medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |