CN110163245A - Class of service prediction technique and system - Google Patents
Class of service prediction technique and system Download PDFInfo
- Publication number
- CN110163245A CN110163245A CN201910274875.3A CN201910274875A CN110163245A CN 110163245 A CN110163245 A CN 110163245A CN 201910274875 A CN201910274875 A CN 201910274875A CN 110163245 A CN110163245 A CN 110163245A
- Authority
- CN
- China
- Prior art keywords
- class
- service
- prediction model
- sample data
- predicted
- 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 description 47
- 230000010354 integration Effects 0.000 claims abstract description 36
- 238000012549 training Methods 0.000 claims description 64
- 230000006870 function Effects 0.000 claims description 19
- 238000012545 processing Methods 0.000 claims description 14
- 239000012141 concentrate Substances 0.000 claims description 4
- 238000013480 data collection Methods 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 16
- 238000005516 engineering process Methods 0.000 description 6
- 238000005070 sampling Methods 0.000 description 6
- 238000004891 communication Methods 0.000 description 5
- 238000013475 authorization Methods 0.000 description 4
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000010276 construction Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 238000009412 basement excavation Methods 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 201000006549 dyspepsia Diseases 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Operations Research (AREA)
- Artificial Intelligence (AREA)
- Tourism & Hospitality (AREA)
- Game Theory and Decision Science (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The disclosure provides the class of service prediction scheme based on hierarchal model.In this scenario, class prediction model has hierarchal model structure, that is, the first class of service prediction model and the second class of service prediction model, the second class of service are the subclass of the first class of service.First class of service prediction model utilizes the first sample data set with the first class of service mark value to train, and the second class of service prediction model utilizes the second sample data set with corresponding second class of service mark value to train.In prediction, the first and second class of service prediction results are predicted using the first and second traffic forecast models respectively, then the first and second class of service prediction results are integrated, and class of service is determined based on the result after integration.It can be realized efficient class of service prediction for the application scenarios between classification with high correlation using the program.
Description
Technical field
The disclosure is usually directed to field of computer technology, more particularly, to for train business category prediction model
Method and device, and the methods, devices and systems of the classification for predicting business datum.
Background technique
With the development of internet, artificial intelligence technology is universal, more and more companies attempt using artificial intelligence or
The relevant technologies of machine learning solve its traffic issues.It is most widely used at present, technology the most mature is two classification
Supervised learning task.For example, given is a large amount of transaction data with mark information, label in the excavation traded extremely
Information is whether the transaction is an abnormal transaction.In task later, future transaction is carried out by building model different
Often transaction possibility prediction, and high abnormal possible transaction is prevented.However, on the other hand, a large amount of learning tasks tool
Have the characteristics that multi-class, that is, possible mark information has a variety of.For example, clicking prediction with the user of the Alipay setting page
For, user may select to click " security setting " (and its sub- option " password setting ", " account authorization ", " unlock be arranged ",
" security centre " etc.), " bio-identification " (and its sub- option), " payment setting " (and its sub- option), " general " (and its son choosing
) etc. different options, it is a variety of to also mean that classification has, rather than only two class usually needs for modeling task
The probability to be clicked for every sub- option prediction user, thus classification sum just becomes more.
Obviously, more classification learning tasks are more increasingly complex and difficult than two classification learning tasks, because it is needed from more
Classification in select prediction classification.In addition, usually having between class imbalance, partial category under industrial application scene
The features such as high correlation.Still it is illustrated so that user's core body of the Alipay setting page is recommended as an example.Firstly, each different
The click volume of option and its sub- option be it is extremely different, for more persons up to several ten million, it is less that few person may only have tens of thousands of minds.Its
Secondary, classification " security setting " can be subdivided into " password setting ", " account authorization ", " unlock setting ", " security centre ", and classification
" bio-identification " can be subdivided into " setting of brush face ", " fingerprint ", " sound lock ".Likewise, for classification " payment is arranged " and " leading to
With " can also further segment.At this point, when to " password setting ", " account authorization ", " unlock is arranged ", " security centre ", brush face
When setting ", " fingerprint ", the sub- option such as " sound lock " establish more disaggregated models, it can be seen that different subitems be have it is big belonging to it
Class, and there is high correlation between the sub- option under the same major class.How under above-mentioned industrial application scene business is carried out
Class prediction becomes urgent problem to be solved.
Summary of the invention
In view of above-mentioned, present disclose provides a kind of class of service prediction technique and devices.Utilize the class of service prediction side
Method and device can realize efficient class of service prediction for the application scenarios between classification with high correlation.
According to one aspect of the disclosure, it provides a kind of for predicting the class method for distinguishing of business datum, comprising: will be to
Prediction business datum is supplied to the first class of service prediction model to carry out the first class of service prediction, to obtain the first service class
Other prediction result；The business datum to be predicted is supplied at least one second class of service prediction model to carry out the second industry
Business class prediction, to obtain at least one second class of service prediction result, the second class of service is time of the first class of service
Grade classification；Obtained first class of service prediction result and at least one second class of service prediction result are carried out at integration
Reason, with the second class of service prediction result after being integrated；And based on the second class of service prediction result after integration, from
The class of service of the business datum to be predicted is determined at least one second class of service, wherein described first and second
Class of service prediction model is multi-class prediction model, the number of the second class of service prediction model and first business
The number of classification is equal and corresponding first class of service of each second class of service prediction model, wherein described first
Class of service prediction model be using with the first class of service mark value first sample data set train and it is described
Second class of service prediction model is trained using the second sample data set with corresponding second class of service mark value
's.
Optionally, in an example of above-mentioned aspect, for training the second sample of the second class of service prediction model
Data set is that the possessed first class of service mark value instruction in the first sample data set belongs to and second business
The sample data set of corresponding first class of service of class prediction model.
Optionally, in an example of above-mentioned aspect, the first class of service prediction result is the industry to be predicted
Business data belong to the probability of each first class of service and the second class of service prediction result is the business to be predicted
Data belong to the probability of each second class of service.
Optionally, in an example of above-mentioned aspect, to obtained first class of service prediction result and the second industry
Business class prediction result carries out integration processing, with the second class of service prediction result after integrate include: be directed to described in extremely
Each second class of service in few second class of service calculates the business datum to be predicted and belongs to second service class
Other probability and the business datum to be predicted belong to the product of the probability of first category corresponding with second class of service, make
Belong to the probability of second class of service for the business datum to be predicted after integration.
Optionally, in an example of above-mentioned aspect, to obtained first class of service prediction result and the second industry
Business class prediction result carries out integration processing, with the second class of service prediction result after integrate include: be directed to described in extremely
Each second class of service in few second class of service, the industry to be predicted after integration is calculated according to predefined function
Business data belong to the probability of second class of service, and the predefined function belongs to second business with the business datum to be predicted
The probability that the probability of classification and the business datum to be predicted belong to first category corresponding with second class of service is from change
Amount.
Optionally, in an example of above-mentioned aspect, for training the first sample of the first class of service prediction model
Notebook data is concentrated, and sample data number possessed by each first class of service meets the first predetermined ratio area requirement；And/or
For training the second sample data of the second class of service prediction model to concentrate, sample possessed by each second class of service
Data number meets the second predetermined ratio area requirement.
According to another aspect of the present disclosure, a kind of method for train business category prediction model is provided, comprising: use
First sample data set with the first class of service mark value trains the first class of service prediction model；And using at least
One the second sample data set trains at least one second class of service prediction model, and the second class of service is the first service class
Other secondary classification, wherein the first and second classs of service prediction model is multi-class prediction model, second business
The number of class prediction model is equal with the number of first class of service, each second class of service prediction model corresponding one
A first class of service and each second class of service prediction model are using with corresponding second class of service mark value
Second sample data set train.
Optionally, in an example of above-mentioned aspect, the first sample data set and second sample data set
It is identical, and for training the sample data of the second class of service prediction model be that possessed first class of service mark value refers to
Show the sample data for belonging to the first class of service corresponding with the second class of service prediction model.
Optionally, in an example of above-mentioned aspect, the method can also include: to for training the first service class
The first sample data set of other prediction model carries out number of samples equilibrium treatment, so that possessed by each first class of service
Sample data number meets the first predetermined ratio area requirement；And/or to for training the second class of service prediction model extremely
Few second sample data set carries out sample data equilibrium treatment, so that each second industry in each second sample data
Sample data number possessed by classification of being engaged in meets the second predetermined ratio area requirement.
According to another aspect of the present disclosure, it provides a kind of for predicting the device of the classification of business datum, comprising: the first industry
Business class prediction unit, is configured as business datum to be predicted being supplied to the first class of service prediction model to carry out the first industry
Business class prediction, to obtain the first class of service prediction result；Second class of service predicting unit, is configured as described to pre-
It surveys business datum and is supplied at least one second class of service prediction model to carry out the second class of service prediction, to obtain at least
One the second class of service prediction result, the second class of service are the secondary classifications of the first class of service；Prediction result integration is single
Member is configured as carrying out obtained first class of service prediction result and at least one second class of service prediction result whole
Conjunction processing, with the second class of service prediction result after being integrated；And class of service determination unit, it is configured as based on whole
The second class of service prediction result after conjunction, determines the business datum to be predicted from least one second class of service
Class of service, wherein the first and second classs of service prediction model is multi-class prediction model, second class of service
The number of prediction model the second class of service prediction model corresponding one equal and each with the number of first class of service
A first class of service, wherein the first class of service prediction model is to utilize the with the first class of service mark value
The second class of service prediction model that the training of one sample data is practised and described is using with corresponding second class of service
What the second sample data set of mark value trained.
Optionally, in an example of above-mentioned aspect, the first class of service prediction result is the industry to be predicted
Business data belong to the probability of each first class of service and the second class of service prediction result is the business to be predicted
Data belong to the probability of each second class of service.
Optionally, in an example of above-mentioned aspect, the prediction result integral unit is configured as: for described in extremely
Each second class of service in few second class of service calculates the business datum to be predicted and belongs to second service class
Other probability and the business datum to be predicted belong to the product of the probability of first category corresponding with second class of service, make
Belong to the probability of second class of service for the business datum to be predicted after integration.
Optionally, in an example of above-mentioned aspect, the prediction result integral unit is configured as: for described in extremely
Each second class of service in few second class of service, the industry to be predicted after integration is calculated according to predefined function
Business data belong to the probability of second class of service, and the predefined function belongs to second business with the business datum to be predicted
The probability that the probability of classification and the business datum to be predicted belong to first category corresponding with second class of service is from change
Amount.
According to another aspect of the present disclosure, a kind of device for train business category prediction model is provided, comprising: first
Training unit is configured with the first sample data set with the first class of service mark value and trains the first class of service
Prediction model；And second training unit, be configured at least one second sample data set train at least one
Two class of service prediction models, the second class of service are the secondary classifications of the first class of service, wherein first and second industry
Business class prediction model is multi-class prediction model, the number of the second class of service prediction model and first service class
Other number is equal, corresponding first class of service of each second class of service prediction model and each second service class
Other prediction model is trained using the second sample data set with corresponding second class of service mark value.
Optionally, in an example of above-mentioned aspect, the first sample data set and second sample data set
It is identical, and for training the sample data of the second class of service prediction model be that possessed first class of service mark value refers to
Show the sample data for belonging to the first class of service corresponding with the second class of service prediction model.
Optionally, in an example of above-mentioned aspect, described device can also include: number of samples equilibrium treatment list
Member is configured as to for training the first sample data set of the first class of service prediction model to carry out at number of samples equilibrium
Reason, so that sample data number possessed by each first class of service meets the requirement of the first predetermined ratio；And/or to being used for
At least one second sample data set of the second class of service prediction model of training carries out sample data equilibrium treatment, so that respectively
Sample data number possessed by each second class of service in a second sample data meets the requirement of the second predetermined ratio.
According to another aspect of the present disclosure, a kind of system for predicting the classification of business datum is provided, comprising: institute as above
The device for the classification for predicting business datum stated；And the dress as described above for being used for train business category prediction model
It sets.
According to another aspect of the present disclosure, a kind of calculating equipment is provided, comprising: at least one processor, and with it is described
The memory of at least one processor coupling, the memory store instruction, when described instruction is by least one described processor
When execution, so that at least one described processor executes as described above for predicting the class method for distinguishing of business datum.
According to another aspect of the present disclosure, a kind of machine readable storage medium is provided, executable instruction is stored with, it is described
Instruction executes the machine as described above for predicting the class method for distinguishing of business datum.
According to another aspect of the present disclosure, a kind of calculating equipment is provided, comprising: at least one processor, and with it is described
The memory of at least one processor coupling, the memory store instruction, when described instruction is by least one described processor
When execution, so that at least one described processor executes the method for being used for train business category prediction model as described above.
According to another aspect of the present disclosure, a kind of machine readable storage medium is provided, executable instruction is stored with, it is described
Instruction executes at least one described processor as described above for train business category prediction model
Method.
Detailed description of the invention
By referring to following attached drawing, may be implemented to further understand the nature and advantages of present disclosure.?
In attached drawing, similar assembly or feature can have identical appended drawing reference.
Fig. 1 shows according to an embodiment of the present disclosure for predicting the block diagram of the system of the classification of business datum；
Fig. 2 shows the schematic diagrames of the hierarchical relationship of class of service according to an embodiment of the present disclosure；
Fig. 3 shows the flow chart of the method according to an embodiment of the present disclosure for train business category prediction model；
Fig. 4 shows the schematic diagram of the method according to an embodiment of the present disclosure for train business category prediction model；
Fig. 5 shows according to an embodiment of the present disclosure for predicting the flow chart of the class method for distinguishing of business datum；
Fig. 6 shows according to an embodiment of the present disclosure for predicting the schematic diagram of the class method for distinguishing of business datum；
Fig. 7 shows the block diagram of class of service prediction model training device according to an embodiment of the present disclosure；
Fig. 8 shows the block diagram of class of service prediction meanss according to an embodiment of the present disclosure；
Fig. 9 shows the box of the calculating equipment according to an embodiment of the present disclosure for train business category prediction model
Figure；
Figure 10 shows according to an embodiment of the present disclosure for predicting the block diagram of the calculating equipment of class of service.
Specific embodiment
Theme described herein is discussed referring now to example embodiment.It should be understood that discussing these embodiments only
It is in order to enable those skilled in the art can better understand that being not to claim to realize theme described herein
Protection scope, applicability or the exemplary limitation illustrated in book.It can be in the protection scope for not departing from present disclosure
In the case of, the function and arrangement of the element discussed are changed.Each example can according to need, omit, substitute or
Add various processes or component.For example, described method can be executed according to described order in a different order, with
And each step can be added, omits or combine.In addition, feature described in relatively some examples is in other examples
It can be combined.
As used in this article, term " includes " and its modification indicate open term, are meant that " including but not limited to ".
Term "based" indicates " being based at least partially on ".Term " one embodiment " and " embodiment " expression " at least one implementation
Example ".Term " another embodiment " expression " at least one other embodiment ".Term " first ", " second " etc. may refer to not
Same or identical object.Here may include other definition, either specific or implicit.Unless bright in context
It really indicates, otherwise the definition of a term is consistent throughout the specification.
For more classification prediction applications, in current method, disaggregated model more than one is constructed usually to be predicted.So
And under industrial application scene, have the characteristics that the high correlation between class imbalance, partial category, usually so as to cause institute
More disaggregated models of building are often difficult to accomplish to take into account each classification when carrying out class of service prediction, and industry thus can occur
Business classification should belong to the low business of sample occurrence rate (that is, less classification of sample data when training), however more classification
Classification that model can recommend sample frequency of occurrence most (that is, sample data when training most classification), so that prediction
Effect is very poor.
A kind of class of service prediction scheme based on hierarchal model is proposed in the disclosure.In this scenario, service class
Other prediction model has hierarchal model structure, that is, class of service prediction model includes the first class of service prediction model and second
Class of service prediction model, wherein the first and second class of service prediction models are multi-class prediction model, the second class of service
It is the secondary classification (subclass) of the first class of service.The number of second class of service prediction model and the number of the first class of service
Mesh is equal and corresponding first class of service of each second class of service prediction model.First class of service prediction model
Be using with the first class of service mark value first sample data set train and the second class of service prediction model
It is to be trained using the second sample data set with corresponding second class of service mark value.In prediction, use respectively
First and second traffic forecast models predict the first and second class of service prediction results, then to the first and second business
Class prediction result is integrated, and determines class of service based on the result after integration.Thereby, it is possible to based on related between classification
Property carry out class of service prediction so that for, with the application scenarios of high correlation, can be realized efficient between classification
Class of service prediction.
In the disclosure, there is characteristic Xi and a variety of for the training dataset D of train business category prediction model
Category label value Yi, yi, that is, D={ (Xi, yi, Yi) }, wherein Yi is the classification mark of the first class of service (that is, category)
Note value, yi are the category label values of the second class of service (that is, second level classification).
It is described in detail below in conjunction with attached drawing according to an embodiment of the present disclosure for predicting the classification of business datum
Method and device and system.
Fig. 1 shows the system according to an embodiment of the present disclosure for predicting the classification of business datum (hereinafter referred to
" class of service forecasting system ") 10 block diagram.As shown in Figure 1, class of service forecasting system 10 includes model training apparatus 100
With class of service prediction meanss 200.
Model training apparatus 100 is configured with sample data and carrys out train business category prediction model.In the disclosure,
Class of service prediction model has hierarchal model structure.Specifically, class of service prediction model is predicted including the first class of service
Model and the second class of service prediction model, wherein the first and second class of service prediction models are multi-class prediction models, and
And second class of service be the first class of service secondary classification (subclass).
Fig. 2 shows the schematic diagrames of the hierarchical relationship of class of service according to an embodiment of the present disclosure.As shown in Fig. 2, one
Grade classification is classification " security setting ", " bio-identification " and " payment setting ".The secondary classification of classification " security setting " is second level
Classification " password setting ", " account authorization ", " unlock setting " and " security centre ", the secondary classification of classification " bio-identification " is two
" the secondary classification of the setting of brush face, " fingerprint " and " sound lock " and classification " pay and be arranged " is that second level classification " is withholdd to grade classification
Sequentially ", " exempt from close payment " and " biology payment ".
In the class of service prediction model structure according to the disclosure, the number and first of the second class of service prediction model
The number of class of service is equal, and corresponding first class of service of each second class of service prediction model.First business
Class prediction model be using with the first class of service mark value first sample data set train and the second business
Class prediction model is trained using the second sample data set with corresponding second class of service mark value.About mould
Type training device 100 will be structurally and operationally described in detail below in reference to attached drawing.
Class of service prediction meanss 200 are configured to carry out using the first and second class of service prediction models pre-
It surveys, the prediction result of the first and second class of service prediction models is integrated, be then based on the result after integrating to determine
Class of service.About being structurally and operationally described in detail below in reference to attached drawing for class of service prediction meanss 200.
Fig. 3 shows the flow chart of the method according to an embodiment of the present disclosure for train business category prediction model.
As shown in figure 3, in block 310, the is trained using the first sample data set with the first class of service mark value
One class of service prediction model.Specifically, the prediction of the first class of service is trained using sample data set D1={ (Xi, Yi) }
Model.
Then, in a block 320, to train at least one second class of service using at least one second sample data set pre-
Model is surveyed, the second class of service is the secondary classification of the first class of service.Here, at least one second class of service prediction model
Number it is equal with the number of the first class of service, and corresponding first service class of each second class of service prediction model
Not.Second sample data set is the sample data set with corresponding second class of service mark value, that is, sample data set D2=
{ (Xi, yi) }.
It is possessed for training the sample data of the second class of service prediction model in an example of the disclosure
First class of service mark value indicates the sample number for belonging to the first class of service corresponding with the second class of service prediction model
According to.That is, for Yi=1 for the second class of service prediction model, used second sample data set is sample data
Collect the sample data of the Yi=1 in D={ (Xi, yi, Yi) }.
Optionally, in addition, in above-mentioned class of service prediction model training method, the first and second sample datas are being used
Before collection to carry out model training to the first and second class of service prediction models, class of service prediction model can also be made
Sample data set carries out sample data equilibrium treatment.Specifically, to for training the first of the first class of service prediction model
Sample data set carries out number of samples equilibrium treatment, so that sample data number possessed by each first class of service meets
First predetermined ratio area requirement.In addition it is also possible to for train the second class of service prediction model at least one second
Sample data set carries out sample data equilibrium treatment, so that each second class of service in each second sample data is had
Some sample data numbers meet the second predetermined ratio area requirement.For example, can make for training the first class of service
In the first sample data set of prediction model, with number of samples possessed by Maximum sample size purpose classification and there is most sample
The ratio of sample data possessed by the classification of this number is in the first predetermined ratio range, and/or makes for training second
Second sample data of class of service prediction model is concentrated, and has number of samples possessed by Maximum sample size purpose classification and tool
There is the ratio of sample data possessed by the classification of smallest sample number in the second predetermined ratio range.Here, the first predetermined ratio
Example area requirement and the second predetermined ratio area requirement can be according to concrete application scene and be arranged, for example, the first predetermined ratio
Example range and the second predetermined ratio range can be set to 1:50.
For example, for the first sample data set D1={ (Xi, Yi) } for training the first class of service prediction model,
When number of samples in first sample data set is unbalanced, that is, there is number of samples possessed by Maximum sample size purpose classification
Ratio with sample data possessed by the classification with smallest sample number is not within the scope of the first predetermined ratio, then to high frequency
The sample (that is, sample data more sample) of Yi carries out down-sampling, so that having Maximum sample size purpose classification to be had
Some numbers of samples with smallest sample number classification possessed by sample data ratio in the first predetermined ratio range
It is interior.
Equally, for for training the second sample data set D2={ (Xi, yi) } of the second class of service prediction model,
When the number of samples that second sample data is concentrated is unbalanced, that is, there is number of samples possessed by Maximum sample size purpose classification
Ratio with sample data possessed by the classification with smallest sample number is not within the scope of the second predetermined ratio, then to high frequency
The sample (that is, sample data more sample) of yi carries out down-sampling, so that having Maximum sample size purpose classification to be had
Some numbers of samples with smallest sample number classification possessed by sample data ratio in the second predetermined ratio range
It is interior.
Fig. 4 shows the schematic diagram of the method according to an embodiment of the present disclosure for train business category prediction model.
In class of service prediction model training method shown in Fig. 4, first category has 3 second level classifications, and in the first business
Number of samples equilibrium treatment has been carried out to sample data when class prediction model (that is, first-level model) training.
Specifically, extracted from original training data collection D={ (Xi, yi, Yi) } first sample data set D1=(Xi,
Yi down-sampling) }, and to the sample of high frequency Yi (that is, sample data more sample) is carried out, so that having maximum sample
The ratio of number of samples possessed by the classification of number and sample data possessed by the classification with smallest sample number is the
Within the scope of one predetermined ratio.Then, first-level model is trained using by down-sampling treated first sample data set D1.
Then, the second sample data set of Yi=1 is extracted respectively from original training data collection D={ (Xi, yi, Yi) }
D2Yi=1={ (Xi, yi) }, D2Yi=2={ (Xi, yi) } and D2Yi=3={ (Xi, yi) }.Then, D2 is used respectivelyYi=1=(Xi,
Yi) }, D2Yi=2={ (Xi, yi) } and D2Yi=3={ (Xi, yi) } trains corresponding second class of service prediction model.
Here, it is to be noted that, for for training the second sample data set D2 of the second class of service prediction modelYi=1
={ (Xi, yi) }, D2Yi=2={ (Xi, yi) } and D2Yi=3={ (Xi, yi) }, in situation unbalanced there are number of samples,
It is also required to carry out number of samples equilibrium treatment, that is, the sample of high frequency yi (that is, sample data more sample) adopt
Sample.
Using above-mentioned class of service prediction model training method, by the sample for the training of class of service prediction model
Data set carries out down-sampling processing to high frequency samples there are in the unbalanced situation of number of samples, so that being used for service class
The number of samples of other prediction model training is balanced, and thus avoiding the occurrence of leads to service class since trained number of samples is unbalanced
The problem of business datum to be predicted is predicted as most high frequency classification in prediction by other prediction model.
Fig. 5 shows according to an embodiment of the present disclosure for predicting the flow chart of the class method for distinguishing of business datum.
As shown in figure 5, business datum to be predicted is supplied to the first class of service prediction model to carry out in block 510
One class of service prediction, to obtain the first class of service prediction result.Here, the first class of service prediction result can be to pre-
Survey the probability that business datum belongs to each first class of service.For example, the probability for Y=1, Y=2 and Y=3 predicted
It is a1, a2 and a3 respectively.
Then, in block 520, business datum to be predicted is supplied at least one second class of service prediction model to carry out
Second class of service prediction, to obtain at least one second class of service prediction result.Here, the second class of service prediction result
It is the probability that business datum to be predicted belongs to each second class of service.For example, it is assumed that being based on the second traffic forecast model M 2_1
Prediction obtains business datum to be predicted to belong to the probability of y=1, y=2, y=3 and y=4 being respectively b1, b2, b3 and b4.Assuming that base
It predicts to obtain business datum to be predicted in the second traffic forecast model M 2_2 and belongs to the probability of y=5, y=6 and y=7 and be respectively
C1, c2 and c3.Belong to y=8, y=9 and y assuming that predicting to obtain business datum to be predicted based on the second traffic forecast model M 2_3
=10 probability is respectively d1, d2 and d3.
Then, in block 530, obtained first class of service prediction result and the second class of service prediction result are carried out
Integration processing, with the second class of service prediction result after being integrated.
In an example of the disclosure, obtained first class of service prediction result and the second class of service are predicted
As a result carry out integration processing, with the second class of service prediction result after being integrated may include: for it is described at least one
Each second class of service in second class of service, calculate business datum to be predicted belong to the probability of second class of service with
Business datum to be predicted belongs to the product of the probability of first category corresponding with second class of service, as after integration to pre-
Survey the probability that business datum belongs to second class of service.For example, as described above, integration after the second class of service prediction result
Are as follows: business datum to be predicted belongs to y=1, and 2 ... 10 probability is respectively a1*b1, a1*b2, a1*b3, a1*b4, a2*c1, a2*
C2, a2*c3, a3*d1, a3*d2 and a3*d3.
In another example of the disclosure, obtained first class of service prediction result and the second class of service are predicted
As a result carry out integration processing, with the second class of service prediction result after being integrated may include: at least one second
Each second class of service in class of service, the business datum to be predicted after integration is calculated according to predefined function f (P1, P2)
Belong to the probability of second class of service.Here, predefined function f (P1, P2) belongs to second business with business datum to be predicted
The probability P 2 that the probability P 1 and business datum to be predicted of classification belong to first category corresponding with second class of service is from change
Amount.In general, predefined function f increases with the increase of P1 and/P2.
Then, in block 540, based on the second class of service prediction result after integration, from least one the second class of service
In determine the class of service of business datum to be predicted.For example, can be based on the second class of service prediction result after integration, it will
Second class of service of obtained maximum probability is determined as the class of service of business datum to be predicted.
Fig. 6 shows according to an embodiment of the present disclosure for predicting the schematic diagram of the class method for distinguishing of business datum.Such as
It shown in Fig. 6, is predicted respectively using first-level model and three second-level models, obtains a first class of service prediction result
With three the second class of service prediction results.Then, it is predicted based on the first class of service prediction result and the second class of service
As a result class of service prediction result integration is carried out.Then, to be predicted to determine based on the class of service prediction result after integration
The class of service of business datum.
In the class method for distinguishing for predicting business datum according to the disclosure, used class of service prediction model
With hierarchal model structure, that is, class of service prediction model includes that the first class of service prediction model and the second class of service are pre-
Model is surveyed, the second class of service is the secondary classification (subclass) of the first class of service.In prediction, respectively using first and the
Two traffic forecast models predict the first and second class of service prediction results, then predict the first and second classs of service
As a result it is integrated, and class of service is determined based on the result after integration.Thereby, it is possible to be carried out based on correlation between classification
Class of service prediction, so that for, with the application scenarios of high correlation, can be realized efficient class of service between classification
Prediction.
In addition, in the class method for distinguishing for predicting business datum according to the disclosure, due to carrying out class of service
When prediction model training, concentrate in used training sample data there are in the unbalanced situation of number of samples, to high frequency sample
Thus this progress down-sampling processing avoids the occurrence of so that the number of samples for the training of class of service prediction model is balanced
Class of service prediction model is caused to predict business datum to be predicted in prediction since trained number of samples is unbalanced
The problem of for most high frequency classification.
Fig. 7 shows class of service prediction model training device (hereinafter referred to model according to an embodiment of the present disclosure
Training device) 100 block diagram.As shown in fig. 7, model training apparatus 100 includes that the first training unit 110 and the second training are single
Member 120.
First training unit 110 is configured with the training of the first sample data set with the first class of service mark value
First class of service prediction model out.Here, the first class of service prediction model is multi-class prediction model.First training unit
110 operation can be with reference to the operation above with reference to Fig. 3 block 310 described.
Second training unit 120 is configured at least one second sample data set and trains at least one second industry
Business class prediction model, the second class of service is the secondary classification of the first class of service.Here, the second class of service prediction model
It is multi-class prediction model.The number of second class of service prediction model is equal with the number of the first class of service, and each
Second class of service prediction model corresponds to first class of service.The operation of second training unit 120 can refer to be joined above
According to the operation of the block 320 of Fig. 3 description.
In addition, being institute for training the sample data of the second class of service prediction model in an example of the disclosure
The the first class of service mark value instruction having belongs to the first class of service corresponding with the second class of service prediction model
Sample data.
In addition, model training apparatus 100 can also include number of samples equilibrium treatment list in another example of the disclosure
First (not shown).The number of samples equilibrium treatment unit is configured as to for training the of the first class of service prediction model
One sample data set carries out number of samples equilibrium treatment, so that sample data number possessed by each first class of service is full
The first predetermined ratio area requirement of foot；And/or to for training at least one second sample of the second class of service prediction model
Data set carries out sample data equilibrium treatment, so that possessed by each second class of service in each second sample data
Sample data number meets the second predetermined ratio area requirement.
Fig. 8 shows the block diagram of class of service prediction meanss 200 according to an embodiment of the present disclosure.As shown in figure 8, industry
Business class prediction device 200 includes the first class of service predicting unit 210, the second class of service predicting unit 220, prediction result
Integral unit 230 and class of service determination unit 240.
First class of service predicting unit 210 is configured as business datum to be predicted being supplied to the prediction of the first class of service
Model carries out the first class of service prediction, to obtain the first class of service prediction result.First class of service predicting unit 210
Operation can with reference to above with reference to Fig. 5 block 510 described operation.
Second class of service predicting unit 220 is configured as business datum to be predicted being supplied at least one second business
Class prediction model carries out the second class of service prediction, to obtain at least one second class of service prediction result, the second industry
Business classification is the secondary classification of the first class of service.The operation of second class of service predicting unit 220 can with reference to above with reference to
The operation of the block 520 of Fig. 5 description.
Prediction result integral unit 230 is configured as to obtained first class of service prediction result and the second service class
Other prediction result carries out integration processing, with the second class of service prediction result after being integrated.Prediction result integral unit 230
Operation can with reference to above with reference to Fig. 5 block 530 described operation.
Class of service determination unit 240 be configured as based on integration after the second class of service prediction result, from least one
The class of service of the business datum to be predicted is determined in a second class of service.The operation of class of service determination unit 240
It can be with reference to the operation above with reference to Fig. 5 block 540 described.
Above with reference to Fig. 1 to Fig. 8, to the method and device for train business category prediction model according to the disclosure,
For predicting that the embodiment of the method and device of the classification of business datum is described.Model training apparatus and business above
Class prediction device can use hardware realization, can also be realized using the combination of software or hardware and software.
Fig. 9 shows the calculating equipment 900 according to an embodiment of the present disclosure for train business category prediction model
Hardware structure diagram.As shown in figure 9, calculating equipment 900 may include at least one processor 910, memory 920,930 and of memory
Communication interface 940, and at least one processor 910, memory 920, memory 930 and communication interface 940 connect via bus 960
It is connected together.At least one processor 910 executes at least one computer-readable instruction for storing or encoding in memory
(that is, above-mentioned element realized in a software form).
In one embodiment, computer executable instructions are stored in memory, make at least one when implemented
Processor 910: the first class of service prediction mould is trained using the first sample data set with the first class of service mark value
Type；And at least one second class of service prediction model, the second business are trained using at least one second sample data set
Classification is the secondary classification of the first class of service, wherein the first and second classs of service prediction model is multi-class prediction
Model, the number of the second class of service prediction model is equal with the number of first class of service, each second business
Corresponding first class of service of class prediction model and each second class of service prediction model are corresponding using having
What the second sample data set of the second class of service mark value trained.
It should be understood that the computer executable instructions stored in memory make at least one processor when implemented
910 carry out the above various operations and functions described in conjunction with Fig. 1-8 in each embodiment of the disclosure.
Figure 10 shows according to an embodiment of the present disclosure for predicting the calculating equipment 1000 of the classification of business datum
Hardware structure diagram.As shown in Figure 10, calculating equipment 1000 may include at least one processor 1010, memory 1020, memory
1030 and communication interface 1040, and at least one processor 1010, memory 1020, memory 1030 and communication interface 1040 pass through
It is linked together by bus 1060.At least one processor 1010 executes at least one calculating for storing or encoding in memory
Machine readable instruction (that is, above-mentioned element realized in a software form).
In one embodiment, computer executable instructions are stored in memory, make at least one when implemented
Processor 1010: being supplied to the first class of service prediction model for business datum to be predicted to carry out the first class of service prediction,
To obtain the first class of service prediction result；The business datum to be predicted is supplied to the prediction of at least one second class of service
Model carries out the second class of service prediction, and to obtain at least one second class of service prediction result, the second class of service is
The secondary classification of first class of service；It is pre- to obtained first class of service prediction result and at least one second class of service
It surveys result and carries out integration processing, with the second class of service prediction result after being integrated；And based on the second industry after integration
Business class prediction as a result, determine the class of service of the business datum to be predicted from least one second class of service,
In, the first and second classs of service prediction model is multi-class prediction model, the second class of service prediction model
Number corresponding first business of the second class of service prediction model equal and each with the number of first class of service
Classification, wherein the first class of service prediction model is to utilize the first sample data with the first class of service mark value
Collecting training and the second class of service prediction model is to utilize the with corresponding second class of service mark value
What two sample data sets trained.
It should be understood that the computer executable instructions stored in memory make at least one processor when implemented
1010 carry out the above various operations and functions described in conjunction with Fig. 1-8 in each embodiment of the disclosure.
In the disclosure, calculating equipment 900/1000 can include but is not limited to: personal computer, server computer,
Work station, desktop computer, laptop computer, notebook computer, mobile computing device, smart phone, plate calculate
Machine, cellular phone, personal digital assistant (PDA), hand-held device, messaging devices, wearable calculating equipment, consumer electronics are set
It is standby etc..
According to one embodiment, a kind of program product of such as non-transitory machine readable media is provided.Non-transitory
Machine readable media can have instruction (that is, above-mentioned element realized in a software form), which when executed by a machine, makes
It obtains machine and executes the above various operations and functions described in conjunction with Fig. 1-8 in each embodiment of the disclosure.Specifically, Ke Yiti
For being furnished with the system or device of readable storage medium storing program for executing, store on the readable storage medium storing program for executing any in realization above-described embodiment
The software program code of the function of embodiment, and read and execute the computer of the system or device or processor and be stored in
Instruction in the readable storage medium storing program for executing.
According to one embodiment, a kind of program product of such as non-transitory machine readable media is provided.Non-transitory
Machine readable media can have instruction (that is, above-mentioned element realized in a software form), which when executed by a machine, makes
It obtains machine and executes the above various operations and functions described in conjunction with Fig. 1-8 in each embodiment of the disclosure.Specifically, Ke Yiti
For being furnished with the system or device of readable storage medium storing program for executing, store on the readable storage medium storing program for executing any in realization above-described embodiment
The software program code of the function of embodiment, and read and execute the computer of the system or device or processor and be stored in
Instruction in the readable storage medium storing program for executing.
In this case, it is real that any one of above-described embodiment can be achieved in the program code itself read from readable medium
The function of example is applied, therefore the readable storage medium storing program for executing of machine readable code and storage machine readable code constitutes of the invention one
Point.
The embodiment of readable storage medium storing program for executing include floppy disk, hard disk, magneto-optic disk, CD (such as CD-ROM, CD-R, CD-RW,
DVD-ROM, DVD-RAM, DVD-RW, DVD-RW), tape, non-volatile memory card and ROM.It selectively, can be by communication network
Network download program code from server computer or on cloud.
It will be appreciated by those skilled in the art that each embodiment disclosed above can be in the situation without departing from invention essence
Under make various changes and modifications.Therefore, protection scope of the present invention should be defined by the appended claims.
It should be noted that step and unit not all in above-mentioned each process and each system construction drawing is all necessary
, certain step or units can be ignored according to the actual needs.Each step execution sequence be not it is fixed, can be according to need
It is determined.Apparatus structure described in the various embodiments described above can be physical structure, be also possible to logical construction, that is, have
A little units may be realized by same physical entity, be realized alternatively, some units may divide by multiple physical entities, alternatively, can be with
It is realized jointly by certain components in multiple autonomous devices.
In the above various embodiments, hardware cell or module mechanically or can be realized electrically.For example, one
Hardware cell, module or processor may include permanent dedicated circuit or logic (such as special processor, FPGA or
ASIC) corresponding operating is completed.Hardware cell or processor can also include programmable logic or circuit (such as general processor or
Other programmable processors), interim setting can be carried out by software to complete corresponding operating.Concrete implementation mode is (mechanical
Mode or dedicated permanent circuit or the circuit being temporarily arranged) it can be determined based on cost and temporal consideration.
The specific embodiment illustrated above in conjunction with attached drawing describes exemplary embodiment, it is not intended that may be implemented
Or fall into all embodiments of the protection scope of claims." exemplary " meaning of the term used in entire this specification
Taste " be used as example, example or illustration ", be not meant to than other embodiments " preferably " or " there is advantage ".For offer pair
The purpose of the understanding of described technology, specific embodiment include detail.However, it is possible in these no details
In the case of implement these technologies.In some instances, public in order to avoid the concept to described embodiment causes indigestion
The construction and device known is shown in block diagram form.
The foregoing description of present disclosure is provided so that any those of ordinary skill in this field can be realized or make
Use present disclosure.To those skilled in the art, the various modifications carried out to present disclosure are apparent
, also, can also answer generic principles defined herein in the case where not departing from the protection scope of present disclosure
For other modifications.Therefore, present disclosure is not limited to examples described herein and design, but disclosed herein with meeting
Principle and novel features widest scope it is consistent.
Claims (21)
1. a kind of for predicting the class method for distinguishing of business datum, comprising:
Business datum to be predicted is supplied to the first class of service prediction model to carry out the first class of service prediction, to obtain
One class of service prediction result；
The business datum to be predicted is supplied at least one second class of service prediction model to carry out the second class of service
Prediction, to obtain at least one second class of service prediction result, the second class of service is the secondary classification of the first class of service；
Integration processing is carried out to obtained first class of service prediction result and at least one second class of service prediction result,
With the second class of service prediction result after being integrated；And
Based on the second class of service prediction result after integration, determined from least one second class of service described to be predicted
The class of service of business datum,
Wherein, the first and second classs of service prediction model is multi-class prediction model, the second class of service prediction
The number of model the second class of service prediction model equal and each with the number of first class of service corresponding one
One class of service,
Wherein, the first class of service prediction model is to utilize the first sample data set with the first class of service mark value
The second class of service prediction model train and described is to utilize second with corresponding second class of service mark value
What sample data set trained.
2. the method for claim 1, wherein for training the second sample data set of the second class of service prediction model
Be in the first sample data set possessed first class of service mark value instruction belong to it is pre- with second class of service
Survey the sample data set of corresponding first class of service of model.
3. method according to claim 1 or 2, wherein the first class of service prediction result is the business to be predicted
Data belong to the probability of each first class of service and the second class of service prediction result is the business number to be predicted
According to the probability for belonging to each second class of service.
4. method as claimed in claim 3, wherein to obtained first class of service prediction result and the second class of service
Prediction result carries out integration processing, includes: with the second class of service prediction result after being integrated
For each second class of service at least one described second class of service, the business datum category to be predicted is calculated
Belong to first category corresponding with second class of service in the probability of second class of service and the business datum to be predicted
Probability product, the probability of second class of service is belonged to as the business datum to be predicted after integration.
5. method as claimed in claim 3, wherein to obtained first class of service prediction result and the second class of service
Prediction result carries out integration processing, includes: with the second class of service prediction result after being integrated
For each second class of service at least one described second class of service, after calculating integration according to predefined function
The business datum to be predicted belong to the probability of second class of service, the predefined function is with the business datum to be predicted
The probability and the business datum to be predicted for belonging to second class of service belong to the first kind corresponding with second class of service
Other probability is independent variable.
6. the method for claim 1, wherein for training the first sample data of the first class of service prediction model
It concentrates, sample data number possessed by each first class of service meets the requirement of the first predetermined ratio；And/or
For training the second sample data of the second class of service prediction model to concentrate, possessed by each second class of service
Sample data number meets the requirement of the second predetermined ratio.
7. a kind of method for train business category prediction model, comprising:
The first class of service prediction model is trained using the first sample data set with the first class of service mark value；And
At least one second class of service prediction model, the second class of service are trained using at least one second sample data set
It is the secondary classification of the first class of service,
Wherein, the first and second classs of service prediction model is multi-class prediction model, the second class of service prediction
The number of model is equal with the number of first class of service, corresponding first industry of each second class of service prediction model
Business classification and each second class of service prediction model are to utilize the second sample with corresponding second class of service mark value
Notebook data collection trains.
8. the method for claim 7, wherein for training the second sample data set of the second class of service prediction model
Be in the first sample data set possessed first class of service mark value instruction belong to it is pre- with second class of service
Survey the sample data set of corresponding first class of service of model.
9. method as claimed in claim 7 or 8, further includes:
To for train the first class of service prediction model first sample data set carry out number of samples equilibrium treatment so that
Sample data number possessed by each first class of service meets the first predetermined ratio area requirement；And/or
To for training at least one second sample data set of the second class of service prediction model to carry out at sample data equilibrium
It manages, makes a reservation for so that sample data number possessed by each second class of service in each second sample data meets second
Proportional region requirement.
10. a kind of for predicting the device of the classification of business datum, comprising:
First class of service predicting unit is configured as business datum to be predicted being supplied to the first class of service prediction model
The first class of service prediction is carried out, to obtain the first class of service prediction result；
Second class of service predicting unit is configured as the business datum to be predicted being supplied at least one second service class
Other prediction model carries out the second class of service prediction, to obtain at least one second class of service prediction result, the second business
Classification is the secondary classification of the first class of service；
Prediction result integral unit is configured as to obtained first class of service prediction result and at least one second business
Class prediction result carries out integration processing, with the second class of service prediction result after being integrated；And
Class of service determination unit, the second class of service prediction result after being configured as based on integration, from least one second
The class of service of the business datum to be predicted is determined in class of service,
Wherein, the first and second classs of service prediction model is multi-class prediction model, the second class of service prediction
The number of model the second class of service prediction model equal and each with the number of first class of service corresponding one
One class of service,
Wherein, the first class of service prediction model is to utilize the first sample data set with the first class of service mark value
The second class of service prediction model train and described is to utilize second with corresponding second class of service mark value
What sample data set trained.
11. device as claimed in claim 10, wherein the first class of service prediction result is the business number to be predicted
It is the business datum to be predicted according to the probability and the second class of service prediction result that belong to each first class of service
Belong to the probability of each second class of service.
12. device as claimed in claim 11, wherein the prediction result integral unit is configured as:
For each second class of service at least one described second class of service, the business datum category to be predicted is calculated
Belong to first category corresponding with second class of service in the probability of second class of service and the business datum to be predicted
Probability product, the probability of second class of service is belonged to as the business datum to be predicted after integration.
13. device as claimed in claim 11, wherein the prediction result integral unit is configured as:
For each second class of service at least one described second class of service, after calculating integration according to predefined function
The business datum to be predicted belong to the probability of second class of service, the predefined function is with the business datum to be predicted
The probability and the business datum to be predicted for belonging to second class of service belong to the first kind corresponding with second class of service
Other probability is independent variable.
14. a kind of device for train business category prediction model, comprising:
First training unit is configured with the first sample data set with the first class of service mark value and trains first
Class of service prediction model；And
Second training unit is configured at least one second sample data set and trains at least one second class of service
Prediction model, the second class of service are the secondary classifications of the first class of service,
Wherein, the first and second classs of service prediction model is multi-class prediction model, the second class of service prediction
The number of model is equal with the number of first class of service, corresponding first industry of each second class of service prediction model
Business classification and each second class of service prediction model are to utilize the second sample with corresponding second class of service mark value
Notebook data collection trains.
15. device as claimed in claim 14, wherein the first sample data set and the second sample data set phase
It together, and for training the sample data of the second class of service prediction model is possessed first class of service mark value instruction
Belong to the sample data of the first class of service corresponding with the second class of service prediction model.
16. the device as described in claims 14 or 15, further includes:
Number of samples equilibrium treatment unit is configured as to for training the first sample data of the first class of service prediction model
Collection carries out number of samples equilibrium treatment, makes a reservation for so that sample data number possessed by each first class of service meets first
Proportional region requirement；And/or to for training at least one second sample data set of the second class of service prediction model to carry out
Sample data equilibrium treatment, so that sample data number possessed by each second class of service in each second sample data
Mesh meets the second predetermined ratio area requirement.
17. a kind of system for predicting the classification of business datum, comprising:
Device as described in any in claim 10 to 13；And
Device as described in any in claim 14 to 16.
18. a kind of calculating equipment, comprising:
At least one processor, and
The memory coupled at least one described processor, the memory store instruction, when described instruction by it is described at least
When one processor executes, so that at least one described processor executes the method as described in any in claims 1 to 6.
19. a kind of machine readable storage medium, is stored with executable instruction, described instruction makes the machine upon being performed
Execute the method as described in any in claims 1 to 6.
20. a kind of calculating equipment, comprising:
At least one processor, and
The memory coupled at least one described processor, the memory store instruction, when described instruction by it is described at least
When one processor executes, so that at least one described processor executes the method as described in any in claim 7 to 10.
21. a kind of machine readable storage medium, is stored with executable instruction, described instruction makes the machine upon being performed
Execute the method as described in any in claim 7 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910274875.3A CN110163245A (en) | 2019-04-08 | 2019-04-08 | Class of service prediction technique and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910274875.3A CN110163245A (en) | 2019-04-08 | 2019-04-08 | Class of service prediction technique and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110163245A true CN110163245A (en) | 2019-08-23 |
Family
ID=67639197
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910274875.3A Pending CN110163245A (en) | 2019-04-08 | 2019-04-08 | Class of service prediction technique and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110163245A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111026908A (en) * | 2019-12-10 | 2020-04-17 | 腾讯科技（深圳）有限公司 | Song label determination method and device, computer equipment and storage medium |
| CN111783143A (en) * | 2020-07-24 | 2020-10-16 | 支付宝(杭州)信息技术有限公司 | Method, system and system for determining service model use of user data |
| CN113298352A (en) * | 2021-04-28 | 2021-08-24 | 北京网核精策科技管理中心(有限合伙) | Enterprise industry information processing method and device, electronic equipment and readable storage medium |
| US11178069B2 (en) | 2020-03-20 | 2021-11-16 | International Business Machines Corporation | Data-analysis-based class of service management for different web resource sections |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA2504181A1 (en) * | 2004-04-15 | 2005-10-15 | Microsoft Corporation | Verifying relevance between keywords and web site contents |
| CN108875776A (en) * | 2018-05-02 | 2018-11-23 | 北京三快在线科技有限公司 | Model training method and device, business recommended method and apparatus, electronic equipment |
| CN108984618A (en) * | 2018-06-13 | 2018-12-11 | 深圳市商汤科技有限公司 | Data processing method and device, electronic equipment and computer readable storage medium |
| CN109102023A (en) * | 2018-08-14 | 2018-12-28 | 阿里巴巴集团控股有限公司 | A kind of method of generating classification model and device, a kind of data identification method and device |
| US20190102701A1 (en) * | 2017-09-29 | 2019-04-04 | Oracle International Corpoation | Techniques for generating a hierarchical model to identify a class among a plurality of classes |
-
2019
- 2019-04-08 CN CN201910274875.3A patent/CN110163245A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA2504181A1 (en) * | 2004-04-15 | 2005-10-15 | Microsoft Corporation | Verifying relevance between keywords and web site contents |
| US20190102701A1 (en) * | 2017-09-29 | 2019-04-04 | Oracle International Corpoation | Techniques for generating a hierarchical model to identify a class among a plurality of classes |
| CN108875776A (en) * | 2018-05-02 | 2018-11-23 | 北京三快在线科技有限公司 | Model training method and device, business recommended method and apparatus, electronic equipment |
| CN108984618A (en) * | 2018-06-13 | 2018-12-11 | 深圳市商汤科技有限公司 | Data processing method and device, electronic equipment and computer readable storage medium |
| CN109102023A (en) * | 2018-08-14 | 2018-12-28 | 阿里巴巴集团控股有限公司 | A kind of method of generating classification model and device, a kind of data identification method and device |
Non-Patent Citations (1)
| Title |
|---|
| 陆旭: "《文本挖掘中若干关键问题研究》", 合肥：中国科学技术大学出版社, pages: 76 - 85 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111026908A (en) * | 2019-12-10 | 2020-04-17 | 腾讯科技（深圳）有限公司 | Song label determination method and device, computer equipment and storage medium |
| CN111026908B (en) * | 2019-12-10 | 2023-09-08 | 腾讯科技（深圳）有限公司 | Song label determining method, device, computer equipment and storage medium |
| US11178069B2 (en) | 2020-03-20 | 2021-11-16 | International Business Machines Corporation | Data-analysis-based class of service management for different web resource sections |
| CN111783143A (en) * | 2020-07-24 | 2020-10-16 | 支付宝(杭州)信息技术有限公司 | Method, system and system for determining service model use of user data |
| CN111783143B (en) * | 2020-07-24 | 2023-05-09 | 支付宝(杭州)信息技术有限公司 | Method, device and system for determining service model use of user data |
| CN113298352A (en) * | 2021-04-28 | 2021-08-24 | 北京网核精策科技管理中心(有限合伙) | Enterprise industry information processing method and device, electronic equipment and readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Khan et al. | Multi-gcn: Graph convolutional networks for multi-view networks, with applications to global poverty | |
| CN110163245A (en) | Class of service prediction technique and system | |
| US11151573B2 (en) | Intelligent chargeback processing platform | |
| US11157972B2 (en) | Document term recognition and analytics | |
| CN108596616B (en) | User data authenticity analysis method and device, storage medium and electronic equipment | |
| JP2021121922A (en) | Multi-model training method and apparatus based on feature extraction, electronic device, and medium | |
| CN109740620A (en) | Method for building up, device, equipment and the storage medium of crowd portrayal disaggregated model | |
| JP2018526704A (en) | Dynamic topological system and method for efficient billing | |
| JP2019525309A (en) | Method and apparatus for controlling data risk | |
| CN109118053B (en) | Method and device for identifying card stealing risk transaction | |
| CN111325619A (en) | Credit card fraud detection model updating method and device based on joint learning | |
| US11423307B2 (en) | Taxonomy construction via graph-based cross-domain knowledge transfer | |
| CN104321794A (en) | A system and method using multi-dimensional rating to determine an entity's future commercial viability | |
| CN108228622A (en) | The sorting technique and device of traffic issues | |
| CN110046784A (en) | A kind of risk of user's access determines method and device | |
| CN113516480A (en) | Payment risk identification method, device and equipment | |
| CN111639690A (en) | Fraud analysis method, system, medium, and apparatus based on relational graph learning | |
| US20230325950A1 (en) | Updating a document based on transaction analysis | |
| CN107038165B (en) | Service parameter acquisition method and device | |
| CN110197426B (en) | Credit scoring model building method, device and readable storage medium | |
| CN109783381B (en) | Test data generation method, device and system | |
| Upreti et al. | Enhanced algorithmic modelling and architecture in deep reinforcement learning based on wireless communication Fintech technology | |
| CN105608460A (en) | Method and system for fusing multiple classifiers | |
| CN112801784A (en) | Bit currency address mining method and device for digital currency exchange | |
| CN116757476A (en) | Method and device for constructing risk prediction model and method and device for risk prevention and control |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20200929 Address after: English genus Applicant after: Innovative advanced technology Co.,Ltd. Address before: English genus Applicant before: Advanced innovation technology Co.,Ltd. Effective date of registration: 20200929 Address after: English genus Applicant after: Advanced innovation technology Co.,Ltd. Address before: A four-storey 847 mailbox in Grand Cayman Capital Building, British Cayman Islands Applicant before: Alibaba Group Holding Ltd. |
|
| TA01 | Transfer of patent application right | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190823 |
|
| RJ01 | Rejection of invention patent application after publication |