CN112069404A - Commodity information display method, device, equipment and storage medium - Google Patents
Commodity information display method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN112069404A CN112069404A CN202010895345.3A CN202010895345A CN112069404A CN 112069404 A CN112069404 A CN 112069404A CN 202010895345 A CN202010895345 A CN 202010895345A CN 112069404 A CN112069404 A CN 112069404A
- Authority
- CN
- China
- Prior art keywords
- information
- commodity
- user
- target
- model
- 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 52
- 238000012549 training Methods 0.000 claims description 56
- 238000003062 neural network model Methods 0.000 claims description 37
- 239000013598 vector Substances 0.000 claims description 37
- 238000004422 calculation algorithm Methods 0.000 claims description 10
- 238000005070 sampling Methods 0.000 claims description 9
- 238000005295 random walk Methods 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 6
- 238000005457 optimization Methods 0.000 abstract 1
- 230000008569 process Effects 0.000 description 12
- 230000006870 function Effects 0.000 description 11
- 238000006243 chemical reaction Methods 0.000 description 9
- 239000000047 product Substances 0.000 description 7
- 230000003287 optical effect Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 230000009471 action Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000006399 behavior Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 239000013307 optical fiber Substances 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- PCTMTFRHKVHKIS-BMFZQQSSSA-N (1s,3r,4e,6e,8e,10e,12e,14e,16e,18s,19r,20r,21s,25r,27r,30r,31r,33s,35r,37s,38r)-3-[(2r,3s,4s,5s,6r)-4-amino-3,5-dihydroxy-6-methyloxan-2-yl]oxy-19,25,27,30,31,33,35,37-octahydroxy-18,20,21-trimethyl-23-oxo-22,39-dioxabicyclo[33.3.1]nonatriaconta-4,6,8,10 Chemical compound C1C=C2C[C@@H](OS(O)(=O)=O)CC[C@]2(C)[C@@H]2[C@@H]1[C@@H]1CC[C@H]([C@H](C)CCCC(C)C)[C@@]1(C)CC2.O[C@H]1[C@@H](N)[C@H](O)[C@@H](C)O[C@H]1O[C@H]1/C=C/C=C/C=C/C=C/C=C/C=C/C=C/[C@H](C)[C@@H](O)[C@@H](C)[C@H](C)OC(=O)C[C@H](O)C[C@H](O)CC[C@@H](O)[C@H](O)C[C@H](O)C[C@](O)(C[C@H](O)[C@H]2C(O)=O)O[C@H]2C1 PCTMTFRHKVHKIS-BMFZQQSSSA-N 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000008707 rearrangement Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9538—Presentation of query results
-
- 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
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0641—Shopping interfaces
- G06Q30/0643—Graphical representation of items or shoppers
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the invention discloses a method, a device, equipment and a storage medium for displaying commodity information. The method comprises the following steps: acquiring a search word input by a target user and first user information of the target user; generating target commodity information according to the search terms; inputting the first user information into a pre-trained first model to obtain the relation between the target commodity information and the target user; determining the display sequence of the target commodity information according to the relation between the target commodity information and a target user; and displaying the target commodity information according to the display sequence. The embodiment of the invention realizes the optimization of the commodity display sequence so as to improve the purchasing power of users.
Description
Technical Field
The embodiment of the invention relates to the internet technology, in particular to a method, a device, equipment and a storage medium for displaying commodity information.
Background
With the rapid development of internet technology, people are more and more accustomed to selecting and purchasing commodities on the internet.
However, click data generated when a user browses commodities is not well utilized, individual commodities cannot be recommended to the user when the user does not click or the cold start of new commodities on shelves is performed, and a merchant cannot obtain personalized and diversified recommendation sequences by using browsing operation data of the users, so that robustness and service conversion efficiency of a recommendation scene cannot be improved, and the merchant income is reduced.
Disclosure of Invention
The embodiment of the invention provides a method, a device, equipment and a storage medium for displaying commodity information, which are used for optimizing a commodity display sequence to improve the purchasing power of a user.
To achieve the purpose, an embodiment of the present invention provides a method for displaying merchandise information, including:
acquiring a search word input by a target user and first user information of the target user;
generating target commodity information according to the search terms;
inputting the first user information into a pre-trained first model to obtain the relation between the target commodity information and the target user;
determining the display sequence of the target commodity information according to the relation between the target commodity information and a target user;
and displaying the target commodity information according to the display sequence.
Further, the inputting the first user information into a pre-trained first model to obtain the relationship between the target commodity information and the target user includes:
inputting the first user information into a pre-trained first model to obtain the relation between the target user and all commodity information;
and determining the relation between the target commodity information and the target user according to the relation between the target user and all the commodity information.
Further, the training of the first model comprises:
acquiring second user information of a sample user, wherein the second user information comprises user label characteristics and historical data, and the historical data comprises purchase information;
and training a preset linear model based on the user label characteristics and the purchase information.
Further, the historical data further includes a commodity name and an access time, and the training of the preset linear model based on the user tag characteristics and the purchase information includes:
sorting the commodity names according to the access time, and marking the commodity information according to the purchase information to obtain first commodity node information;
carrying out random walk sampling on the first commodity node information based on a depth walk algorithm to obtain a second commodity information node;
acquiring a feature vector of the second commodity node information based on a pre-trained first neural network model;
and training a preset second neural network model based on the feature vector and the purchase information, and training a preset linear model based on the user label feature and the purchase information.
Further, the historical data further includes browsing duration, the sorting the commodity names according to the access time, and the marking the commodity information according to the purchase information to obtain the first commodity node information includes:
sorting the commodity names according to the access time, and marking the commodity information according to the purchase information to obtain third commodity node information;
and removing the third commodity node information with the browsing duration being less than a first threshold value to obtain first commodity node information.
Further, the training a preset second neural network model based on the feature vector and the purchase information, and the training a preset linear model based on the user tag feature and the purchase information includes:
and taking the feature vector and the user label feature as input, taking the purchase information as output, and training a first model, wherein the first model comprises a preset second neural network model, a preset linear model and a preset logic loss function.
Further, the first neural network model is a Word2Vec model, and the second neural network model is a feedforward neural network model.
On one hand, the embodiment of the invention also provides a display device of commodity information, which comprises:
the information acquisition module is used for acquiring search terms input by a target user and first user information of the target user;
the information generation module is used for generating target commodity information according to the search terms;
the relationship generation module is used for inputting the first user information into a first model trained in advance to obtain the relationship between the target commodity information and the target user;
the order determining module is used for determining the display order of the target commodity information according to the relation between the target commodity information and the target user;
and the information display module is used for displaying the target commodity information according to the display sequence.
On the other hand, an embodiment of the present invention further provides a computer device, where the computer device includes: one or more processors; a storage device for storing one or more programs which, when executed by the one or more processors, cause the one or more processors to implement a method as provided by any embodiment of the invention.
In yet another aspect, the embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and the computer program, when executed by a processor, implements the method provided in any embodiment of the present invention.
The method comprises the steps of acquiring search terms input by a target user and first user information of the target user; generating target commodity information according to the search terms; inputting the first user information into a pre-trained first model to obtain the relation between the target commodity information and the target user; determining the display sequence of the target commodity information according to the relation between the target commodity information and a target user; the target commodity information is displayed according to the display sequence, the problem that click data generated when a user browses commodities is not utilized, robustness of a recommended scene cannot be improved, business conversion efficiency cannot be improved, and accordingly merchant income is reduced is solved, and the effect that the commodity display sequence is optimized to improve user purchasing power is achieved.
Drawings
Fig. 1 is a schematic flow chart of a method for displaying merchandise information according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating a training method of a first model according to a second embodiment of the present invention;
fig. 3 is a schematic structural diagram of a directed weighted graph generated based on first commodity node information according to a second embodiment of the present invention;
fig. 4 is a flowchart illustrating a method for displaying merchandise information according to a second embodiment of the present invention;
fig. 5 is a schematic structural diagram of a display device for merchandise information according to a third embodiment of the present invention;
fig. 6 is a schematic structural diagram of a computer device according to a fourth embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are for purposes of illustration and not limitation. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
Before discussing exemplary embodiments in more detail, it should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although a flowchart may describe the steps as a sequential process, many of the steps can be performed in parallel, concurrently or simultaneously. In addition, the order of the steps may be rearranged. A process may be terminated when its operations are completed, but may have additional steps not included in the figure. A process may correspond to a method, a function, a procedure, a subroutine, a subprogram, etc.
Furthermore, the terms "first," "second," and the like may be used herein to describe various orientations, actions, steps, elements, or the like, but the orientations, actions, steps, or elements are not limited by these terms. These terms are only used to distinguish one direction, action, step or element from another direction, action, step or element. For example, a first module may be termed a second module, and, similarly, a second module may be termed a first module, without departing from the scope of the present application. The first module and the second module are both modules, but they are not the same module. The terms "first", "second", etc. are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, features defined as "first", "second", may explicitly or implicitly include one or more of the described features. In the description of the embodiments of the present invention, "a plurality" means at least two, e.g., two, three, etc., unless specifically limited otherwise.
Example one
As shown in fig. 1, an embodiment of the present invention provides a method for displaying commodity information, where the method includes:
s110, obtaining the search terms input by the target user and the first user information of the target user.
In this embodiment, if a target user wants to purchase a certain commodity in a certain program, a corresponding search term is input, and at this time, the search term input by the target user and first user information of the target user are obtained, where the program requires the target user to input the first user information of the target user, and after the target user logs in the program, the first user information input by the target user and the first user information generated by the target user using the program can be obtained.
And S120, generating target commodity information according to the search terms.
S130, inputting the first user information into a first model trained in advance to obtain the relation between the target commodity information and the target user.
S140, determining the display sequence of the target commodity information according to the relation between the target commodity information and the target user.
And S150, displaying the target commodity information according to the display sequence.
In this embodiment, after the search term input by the target user is obtained, the target commodity information needs to be generated directly according to the search term, that is, the target commodity information corresponding to the search term can be generated after the search term is input by the target user, but the target commodity information generated by light cannot meet the needs of merchants, and therefore, a display sequence of the target commodity information for the target user needs to be generated, so as to improve the purchase desire and user experience of the target user.
Specifically, the first user information acquired previously is input to the first model, wherein the first model is pre-trained by using a large amount of user information, so that the relationship between the target commodity information and the target user can be obtained by inputting the first user information to the first model, that is, the preference degree of the target user for each target commodity information is obtained, and thus, the display sequence of the target commodity information can be determined according to the relationship between the target commodity information and the target user, for example, the target commodity information which is more preferred by the target user is more advanced in the display sequence, and finally, the target commodity information is displayed according to the display sequence. Therefore, after the target user searches, the target user can quickly find the interested commodity information, the purchase desire of the target user is greatly improved, and the business conversion rate and the merchant income are also improved.
The method comprises the steps of acquiring search terms input by a target user and first user information of the target user; generating target commodity information according to the search terms; inputting the first user information into a pre-trained first model to obtain the relation between the target commodity information and the target user; determining the display sequence of the target commodity information according to the relation between the target commodity information and a target user; the target commodity information is displayed according to the display sequence, the problem that click data generated when a user browses commodities is not utilized, robustness of a recommended scene cannot be improved, business conversion efficiency cannot be improved, and accordingly merchant income is reduced is solved, and the effect that the commodity display sequence is optimized to improve user purchasing power is achieved.
Example two
As shown in fig. 2 to fig. 4, a second embodiment of the present invention provides a method for displaying commodity information, and the second embodiment of the present invention further explains and explains on the basis of the first embodiment of the present invention, and before executing the method for displaying commodity information, training of the first model needs to be performed, and as shown in fig. 2, the specific training method may be:
s210, second user information of the sample user is obtained, wherein the second user information comprises user label characteristics and historical data, and the historical data comprises purchase information, browsing duration, commodity names and access time.
In this embodiment, second user information of a plurality of sample users is obtained for training, and when a sample user is specified, the second user information of the sample user is obtained, where the second user information includes a user tag feature and historical data, and the historical data includes purchase information, browsing duration, a commodity name, and access time.
For example, the user tag features personal information input by the sample user, such as sex, age and preference selected by the user, the historical data is past operation data of the sample user, and a small part of the historical data may be as shown in table 1:
TABLE 1
| Time of access | Duration of browsing | Name of commodity | Purchasing information |
| 2020/1/1 14:20 | 20 seconds | A | Not purchasedBuy |
| 2020/1/1 14:21 | 10 seconds | B | Not purchased |
| 2020/1/1 14:22 | 1 second | C | Not purchased |
| 2020/1/2 14:20 | 5 seconds | D | Not purchased |
| 2020/1/2 14:21 | 30 seconds | E | Purchasing |
S220, sequencing the commodity names according to the access time, and marking the commodity information according to the purchase information to obtain third commodity node information.
In this embodiment, after the historical data is obtained, the commodity names need to be sorted according to the access time, and the commodity information needs to be marked according to the purchase information to obtain third commodity node information, so as to complete preprocessing on the historical data. Specifically, after the historical data as shown in table 1 is obtained, the names of the commodities are sorted according to the access time, and then: and (3) ABCDE, marking the commodity information according to the purchase information, and considering that the purchased commodity information is in a strong relationship, namely: ABCD "E", in addition, for a behavior of a user for multiple days, the behavior is divided by day, because a jump of more than 1 day is large, the effect of the first model is affected, and finally, the obtained third commodity node information is: ABC, D "E".
S230, the third commodity node information with the browsing duration being smaller than the first threshold is removed to obtain first commodity node information.
Further, the browsing time is very short and may be a false touch of the sample user, and the third commodity node information with the browsing time being less than the first threshold value needs to be removed to obtain the first commodity node information. Illustratively, the first threshold is 3 seconds, and thus the first commodity node information is obtained as: AB, D "E".
S240, carrying out random walk sampling on the first commodity node information based on a depth walk algorithm to obtain a second commodity information node.
In this embodiment, after the first commodity node information is obtained, the first commodity node information may be subjected to random Walk sampling based on a Deep Walk algorithm (Deep Walk) to obtain a second commodity information node. The depth Walk algorithm is a depth-first traversal algorithm for repeatedly accessing an accessed node, and the process is repeated until the length of an access sequence meets a preset condition. In this embodiment, preferably, in order to increase the influence of the connection conversion, the connection is used as a global context, and the weights of the connection and the nearest node are increased.
Illustratively, as shown in fig. 3, E is a connecting node, so that a to E, B to E, and D to E are all increased by a conversion weight, all 1, because D to E is nearest, the conversion weight is set to 3 here to enhance the weight of the adjacent sequence. The sequences are extended to a homogeneous network through transformation to construct a directed weighted graph. Further, random walk sampling can be performed from each node in the directed weighted graph to obtain locally associated training data, and a new second commodity information node is constructed: ABDE; BED; AE; BE; and (DE). Therefore, the problems of no click of individual commodity information and cold start of newly-placed commodity information can be effectively solved, more possibilities can be learned by the first model in advance, and the robustness of the first model is enhanced. Preferably, when the data volume is large, the process of the random walk sampling node is slow, path sampling can be accelerated in a parallel mode, when multiple processes are adopted for acceleration, compared with the case that one process pool is opened, one process is started in an outer loop every time, num _ walks with the designated number are distributed to each process in a fixed mode, and therefore the time overhead of frequent creation and destruction of the processes can be reduced to the maximum extent.
And S250, acquiring a feature vector of the second commodity node information based on a pre-trained first neural network model.
In this embodiment, after the second commodity node information is obtained, the feature vector of the second commodity node information may be obtained based on the pre-trained first neural network model. The first neural network model is a Word2Vec model, specifically a Skip-Gram model, wherein the Skip-Gram model can also be trained in advance by adopting data of a sample user, and the purpose of the training is to obtain a weight matrix of a hidden layer in the Skip-Gram model. Specifically, given a specific commodity node information as an input, and then randomly selecting a commodity node information from the vicinity of the commodity node information, the Skip-Gram model outputs a probability that each commodity node information of all the commodity information appears within a window size, which may be preset by a developer, and the output probability is related to a possibility of finding each other commodity node information in the vicinity of the input commodity node information. Optionally, the Skip-Gram model is trained by using word pair (output word) in the specified window size in the text, that is, the input of the Skip-Gram model is selected as the second commodity node information, and the output is the corresponding probability. Illustratively, the input is ABDE, the window size is chosen to be 2, then the training samples are (A, B), (A, D), (B, A), (B, D), (B, E), (D, B), (D, A), (D, E), (E, B) and (E, D), and the output of the input is a vector matrix composed of the probability of occurrence corresponding to each of the above samples.
For example, because the neural network model only accepts numerical input, each commodity node information cannot be directly input into a neural network, and therefore, a commodity table needs to be constructed for all second commodity node information, and then each commodity node information in the commodity table is represented in a One-Hot (One-Hot) coding manner. For example, a commodity table with a size of 10000 is available, an One-Hot vector needs to be constructed for each commodity node information, the position of the current commodity node information corresponding to each commodity node information is required to be 1, and all other positions are required to be 0, so that 10000 vectors with a length of 10000 can be obtained, wherein only One position of each vector is 1. The output of the Skip-Gram model is a 10000-dimensional vector, and represents the prediction probability that all the product node information appears in the vicinity of the input product node information in the product table with respect to the input second product node information. After the training of the first neural network model is completed, each commodity node information has a corresponding vector representation, so that a feature vector of the second commodity node information can be obtained, and the feature vector is an Embedding feature vector.
S260, taking the feature vector and the user label feature as input, taking the purchase information as output, and training a first model, wherein the first model comprises a preset second neural network model, a preset linear model and a preset logic loss function.
In this embodiment, after obtaining the feature vector of the second commodity node information, the feature vector and the user tag feature may be used as input, and the purchase information may be used as output, to train the first model, where the first model includes a preset second neural network model, a preset linear model, and a preset logic loss function, and as an optimal choice, the second neural network model is a feed-forward neural network model.
Specifically, the first model is regarded as a wide & deep model for training, a preset linear model is arranged in the wide part and is used for memorizing, and the user label characteristics and the purchase information are used for training. The deep part is a preset feedforward neural network model and is mainly responsible for generalization, the characteristic vectors and the purchase information are used for training, the similarity relation among some implicit commodity information can be calculated, and the recommendation diversity is improved.
Illustratively, the wide portion is a simple linear model: y — wx + b, where y is the prediction target, x — x [ x1, x 2.., xd ], x is the vector of d features (features), w — w1, w 2.., wd ], w is the parameter of the linear model, and b is the bias (bias) parameter. Wherein the d features include the original input feature and the converted feature. Preferably, the features used are converted into a cross-product (cross-product) conversion. For example, x1 is gender, x1 ═ 0 represents male, x1 ═ 1 represents female, x2 is hobby, x2 ═ 0 represents disliked kendiryl, and x2 ═ 1 represents liking of kendiryl, then new features can be constructed using x1 and x2, x3 ═ (x1& & x2), x3 ═ 1 represents female and likes kendiryl, if not female or disliked kendiryl, x3 ═ 0 and x3 are transformed features, so as to obtain the influence of cross features on the prediction target, and add nonlinearity to the linear model, thereby training the preset linear model.
Further, the deep part is a feedforward neural network model, the feature vectors of the acquired high-dimensional sparse second commodity node information can be converted into low-dimensional dense feature vectors, and then the feature vectors are used as the input of a hidden layer (hidden layers) of the feedforward neural network model for training. The calculation formula of the hidden layer is as follows:
a(l+1)＝f(W(l)a(l)+b(l))
where f is the activation function (e.g., ReLu), a is the last output of the hidden layer, W is the parameter to be trained, and b is the bias parameter.
Finally, the outputs of the wide and deep portions are combined by weight sum and then jointly trained by a logistic loss function (logistic loss). For the window portion, the FTRL (Follow-the-regularized-Leader) algorithm is typically used for training. And for the deep part, an AdaGrad (the learning rate of each dimension can be adjusted according to the gradient value of the independent variable in each dimension, so that the problem that the unified dimension is difficult to adapt to all dimensions) algorithm is adopted for training. The trained first model can determine that the commodity information of Kenday and the target user are related, and can also determine that the commodity information of McDonald and the target user have certain relevance, exemplarily, for the user with the preference of Kenday.
Specifically, the wide & deep model, i.e. the first model, connects the wide part of the single input layer with the deep part composed of the Embedding feature vector layer and the multiple hidden layers, and inputs the wide part and the deep part together into the final output layer. And finally, combining the wide part and the deep part by using a logistic regression model through an output layer to form a unified model. Where the wide and deep portions use a weighted sum of their output log probabilities as predictions, which are then input to a common logic loss function of the joint training. Compared with ensemble learning, joint training simultaneously considers the model of the wide part and the model of the deep part and the weighted sum to optimize all parameters during training, the wide part only needs to supplement the deficiency of the model of the deep part through a small number of feature transforms, and the size of the whole first model is relatively small. The finally obtained first model takes the feature vector and the user label feature as input, the purchase information is taken as output for training, the feature vector is input into the feedforward neural network model in the actual training process, the user label feature is input into the linear model, and the final output is set as the purchase information, so that the relation between the sample user and the commodity information can be judged, for example, the relation is judged according to the user label feature, and the sample user prefers the commodity information by combining the more clicks (the times of appearance of the commodity name), the more purchases times and the longer browsing time, which is equivalent to training the relation between the user label feature, the historical data and whether the sample user can purchase the commodity. According to the trained first model, first user information of a target user is input, the first user information is converted to obtain a feature vector, and then the relation between the target user and all commodity information can be obtained, wherein all commodity information is all commodity information clicked or purchased by a sample user.
It should be noted that the first model may also be only a preset linear model, the second user information obtained at this time only needs to include the user tag feature and the purchase information, the first model may also be only a preset second neural network model, the second user information obtained at this time only needs to include the commodity name, the access time, the browsing duration and the purchase information, and the developer may select training of the model based on the integrity of the second user information of the sample user. Preferably, in order to further ensure the accuracy, even in the case that the second user information is missing, an average number or a median may be taken to fill in the missing information, and the training of the first model is completed by using a preset second neural network model, a preset linear model and a preset logic loss function. And the information type of the second user information used for training is consistent with the information type of the first user information of the target user acquired under the actual condition.
Further, as shown in fig. 4, after the training of the first model, a method for displaying commodity information provided by the second embodiment of the present invention may be performed, where the method includes:
s310, obtaining the search terms input by the target user and the first user information of the target user.
In this embodiment, because the second user information is used in the training of the first model, the first user information also needs to be acquired, where the first user information is the same as the second user information, and the difference is only that the objects are different, so that the first user information includes the user tag characteristics of the target user and the historical data of the target user, where the historical data includes purchase information, browsing duration, a product name, and access time. The content specifically included in the first user information depends on the content included in the second user information of the sample user obtained during the training of the first model, and the content of the content is kept consistent.
And S320, generating target commodity information according to the search terms.
S330, inputting the first user information into a first model trained in advance to obtain the relation between the target user and all commodity information.
In this embodiment, if the second user information is converted into the feature vector during the training of the first model and the second neural network model is used to participate in the training, the first user information also needs to be converted in this step, specifically, the commodity names are sorted according to the access time, and the commodity information is marked according to the purchase information to obtain third commodity node information; removing third commodity node information with the browsing time being less than a first threshold value to obtain first commodity node information; carrying out random walk sampling on the first commodity node information based on a depth walk algorithm to obtain a second commodity information node; and obtaining a feature vector of the second commodity node information based on a pre-trained first neural network model, and finally inputting the feature vector and the user label features into the first model to obtain the relation between the target user and all commodity information.
S340, determining the relation between the target commodity information and the target user according to the relation between the target user and all the commodity information.
In this embodiment, after the first user information is input into the pre-trained first model, the obtained relationship may be the relationship between the target user and all the commodity information, and therefore, the relationship between the target commodity information and the target user needs to be determined according to the relationship between the target user and all the commodity information, that is, the target commodity information is found in all the commodity information, and then the relationship between the target commodity information and the target user is obtained.
And S350, determining the display sequence of the target commodity information according to the relation between the target commodity information and the target user.
And S360, displaying the target commodity information according to the display sequence.
In this embodiment, the implementation method between the steps may refer to a training method of the first model.
According to the embodiment of the invention, through training the first model, the problem that the robustness of a recommended scene and the business conversion efficiency cannot be improved and the merchant income is reduced due to the fact that click data generated when a user browses commodities is not utilized is solved, and the effect of optimizing the commodity display sequence to improve the purchasing power of the user is realized.
EXAMPLE III
As shown in fig. 5, a third embodiment of the present invention provides a display apparatus 100 for merchandise information, where the display apparatus 100 for merchandise information provided by the third embodiment of the present invention can execute a method for displaying merchandise information provided by any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method. The merchandise information display device 100 includes an information acquisition module 200, an information generation module 300, a relationship generation module 400, an order determination module 500, and an information display module 600.
Specifically, the information obtaining module 200 is configured to obtain a search term input by a target user and first user information of the target user; the information generating module 300 is configured to generate target commodity information according to the search terms; the relationship generation module 400 is configured to input the first user information into a first model trained in advance to obtain a relationship between the target commodity information and a target user; the order determining module 500 is configured to determine a display order of the target commodity information according to a relationship between the target commodity information and a target user; the information display module 600 is configured to display the target commodity information according to the display sequence.
In this embodiment, the relationship generating module 400 is specifically configured to input the first user information into a first model trained in advance to obtain the relationship between the target user and all the commodity information; and determining the relation between the target commodity information and the target user according to the relation between the target user and all the commodity information.
Further, the display apparatus 100 for the commodity information further includes a model training module 700, where the model training module 700 is configured to obtain second user information of the sample user, where the second user information includes a user tag feature and historical data, and the historical data includes purchase information; and training a preset linear model based on the user label characteristics and the purchase information.
Further, the historical data further includes commodity names and access time, and the model training module 700 is specifically configured to sort the commodity names according to the access time, and mark the commodity information according to the purchase information to obtain first commodity node information; carrying out random walk sampling on the first commodity node information based on a depth walk algorithm to obtain a second commodity information node; acquiring a feature vector of the second commodity node information based on a pre-trained first neural network model; and training a preset second neural network model based on the feature vector and the purchase information, and training a preset linear model based on the user label feature and the purchase information.
Further, the historical data further includes browsing duration, and the model training module 700 is further specifically configured to sort the commodity names according to the access time, and mark the commodity information according to the purchase information to obtain third commodity node information; and removing the third commodity node information with the browsing duration being less than a first threshold value to obtain first commodity node information.
Further, the model training module 700 is specifically configured to train a first model by using the feature vector and the user tag feature as inputs and the purchase information as outputs, where the first model includes a preset second neural network model, a preset linear model, and a preset logic loss function. The first neural network model is a Word2Vec model, and the second neural network model is a feedforward neural network model.
Example four
Fig. 6 is a schematic structural diagram of a computer device 12 according to a fourth embodiment of the present invention. FIG. 6 illustrates a block diagram of an exemplary computer device 12 suitable for use in implementing embodiments of the present invention. The computer device 12 shown in FIG. 6 is only an example and should not bring any limitations to the functionality or scope of use of embodiments of the present invention.
As shown in FIG. 6, computer device 12 is in the form of a general purpose computing device. The components of computer device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 that couples various system components including the system memory 28 and the processing unit 16.
The system memory 28 may include computer system readable media in the form of volatile memory, such as Random Access Memory (RAM)30 and/or cache memory 32. Computer device 12 may further include other removable/non-removable, volatile/nonvolatile computer system storage media. By way of example only, storage system 34 may be used to read from and write to non-removable, nonvolatile magnetic media (not shown in FIG. 6, and commonly referred to as a "hard drive"). Although not shown in FIG. 6, a magnetic disk drive for reading from and writing to a removable, nonvolatile magnetic disk (e.g., a "floppy disk") and an optical disk drive for reading from or writing to a removable, nonvolatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 18 by one or more data media interfaces. Memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to carry out the functions of embodiments of the invention.
A program/utility 40 having a set (at least one) of program modules 42 may be stored, for example, in memory 28, such program modules 42 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which examples or some combination thereof may comprise an implementation of a network environment. Program modules 42 generally carry out the functions and/or methodologies of the described embodiments of the invention.
The processing unit 16 executes various functional applications and data processing by executing programs stored in the system memory 28, for example, implementing the methods provided by the embodiments of the present invention:
acquiring a search word input by a target user and first user information of the target user;
generating target commodity information according to the search terms;
inputting the first user information into a pre-trained first model to obtain the relation between the target commodity information and the target user;
determining the display sequence of the target commodity information according to the relation between the target commodity information and a target user;
and displaying the target commodity information according to the display sequence.
EXAMPLE five
The fifth embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the methods provided in all the embodiments of the present invention of the present application:
acquiring a search word input by a target user and first user information of the target user;
generating target commodity information according to the search terms;
inputting the first user information into a pre-trained first model to obtain the relation between the target commodity information and the target user;
determining the display sequence of the target commodity information according to the relation between the target commodity information and a target user;
and displaying the target commodity information according to the display sequence.
Computer storage media for embodiments of the invention may employ any combination of one or more computer-readable media. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: 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 the context of this document, 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.
A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. 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 wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Computer program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C + + or the like 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 type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims (10)
1. A commodity information display method is characterized by comprising the following steps:
acquiring a search word input by a target user and first user information of the target user;
generating target commodity information according to the search terms;
inputting the first user information into a pre-trained first model to obtain the relation between the target commodity information and the target user;
determining the display sequence of the target commodity information according to the relation between the target commodity information and a target user;
and displaying the target commodity information according to the display sequence.
2. The method of claim 1, wherein the inputting the first user information into a pre-trained first model to obtain the relationship between the target commodity information and the target user comprises:
inputting the first user information into a pre-trained first model to obtain the relation between the target user and all commodity information;
and determining the relation between the target commodity information and the target user according to the relation between the target user and all the commodity information.
3. The method of claim 1, wherein the training of the first model comprises:
acquiring second user information of a sample user, wherein the second user information comprises user label characteristics and historical data, and the historical data comprises purchase information;
and training a preset linear model based on the user label characteristics and the purchase information.
4. The method of claim 3, wherein the historical data further comprises a commodity name and an access time, and wherein training a preset linear model based on the user tag characteristics and purchase information comprises:
sorting the commodity names according to the access time, and marking the commodity information according to the purchase information to obtain first commodity node information;
carrying out random walk sampling on the first commodity node information based on a depth walk algorithm to obtain a second commodity information node;
acquiring a feature vector of the second commodity node information based on a pre-trained first neural network model;
and training a preset second neural network model based on the feature vector and the purchase information, and training a preset linear model based on the user label feature and the purchase information.
5. The method of claim 4, wherein the historical data further comprises browsing duration, and the sorting the commodity names according to the access time and marking the commodity information according to the purchase information to obtain the first commodity node information comprises:
sorting the commodity names according to the access time, and marking the commodity information according to the purchase information to obtain third commodity node information;
and removing the third commodity node information with the browsing duration being less than a first threshold value to obtain first commodity node information.
6. The method of claim 4, wherein training a preset second neural network model based on the feature vectors and purchase information, and training a preset linear model based on the user tag features and purchase information comprises:
and taking the feature vector and the user label feature as input, taking the purchase information as output, and training a first model, wherein the first model comprises a preset second neural network model, a preset linear model and a preset logic loss function.
7. The method of claim 4, wherein the first neural network model is a Word2Vec model and the second neural network model is a feed-forward neural network model.
8. A display device of commodity information is characterized by comprising:
the information acquisition module is used for acquiring search terms input by a target user and first user information of the target user;
the information generation module is used for generating target commodity information according to the search terms;
the relationship generation module is used for inputting the first user information into a first model trained in advance to obtain the relationship between the target commodity information and the target user;
the order determining module is used for determining the display order of the target commodity information according to the relation between the target commodity information and the target user;
and the information display module is used for displaying the target commodity information according to the display sequence.
9. A computer device, comprising:
one or more processors;
a storage device for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-7.
10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the method according to any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010895345.3A CN112069404A (en) | 2020-08-31 | 2020-08-31 | Commodity information display method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010895345.3A CN112069404A (en) | 2020-08-31 | 2020-08-31 | Commodity information display method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112069404A true CN112069404A (en) | 2020-12-11 |
Family
ID=73664856
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010895345.3A Pending CN112069404A (en) | 2020-08-31 | 2020-08-31 | Commodity information display method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112069404A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113781152A (en) * | 2021-02-08 | 2021-12-10 | 北京沃东天骏信息技术有限公司 | Object recommendation method and device |
| CN115545847A (en) * | 2022-10-21 | 2022-12-30 | 深圳市凯盛浩科技有限公司 | Commodity identification and search method and system based on video stream |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102591876A (en) * | 2011-01-14 | 2012-07-18 | 阿里巴巴集团控股有限公司 | Sequencing method and device of search results |
| WO2016052149A1 (en) * | 2014-09-29 | 2016-04-07 | 富士フイルム株式会社 | Commodity recommendation device and commodity recommendation method |
| CN106127525A (en) * | 2016-06-27 | 2016-11-16 | 浙江大学 | A kind of TV shopping Method of Commodity Recommendation based on sorting algorithm |
| CN107563859A (en) * | 2017-08-31 | 2018-01-09 | 深圳前海弘稼科技有限公司 | Method of Commodity Recommendation, device, equipment and storage medium |
| CN108182621A (en) * | 2017-12-07 | 2018-06-19 | 合肥美的智能科技有限公司 | The Method of Commodity Recommendation and device for recommending the commodity, equipment and storage medium |
| CN110910199A (en) * | 2019-10-16 | 2020-03-24 | 中国平安人寿保险股份有限公司 | Item information sorting method and device, computer equipment and storage medium |
-
2020
- 2020-08-31 CN CN202010895345.3A patent/CN112069404A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102591876A (en) * | 2011-01-14 | 2012-07-18 | 阿里巴巴集团控股有限公司 | Sequencing method and device of search results |
| WO2016052149A1 (en) * | 2014-09-29 | 2016-04-07 | 富士フイルム株式会社 | Commodity recommendation device and commodity recommendation method |
| CN106127525A (en) * | 2016-06-27 | 2016-11-16 | 浙江大学 | A kind of TV shopping Method of Commodity Recommendation based on sorting algorithm |
| CN107563859A (en) * | 2017-08-31 | 2018-01-09 | 深圳前海弘稼科技有限公司 | Method of Commodity Recommendation, device, equipment and storage medium |
| CN108182621A (en) * | 2017-12-07 | 2018-06-19 | 合肥美的智能科技有限公司 | The Method of Commodity Recommendation and device for recommending the commodity, equipment and storage medium |
| CN110910199A (en) * | 2019-10-16 | 2020-03-24 | 中国平安人寿保险股份有限公司 | Item information sorting method and device, computer equipment and storage medium |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113781152A (en) * | 2021-02-08 | 2021-12-10 | 北京沃东天骏信息技术有限公司 | Object recommendation method and device |
| CN115545847A (en) * | 2022-10-21 | 2022-12-30 | 深圳市凯盛浩科技有限公司 | Commodity identification and search method and system based on video stream |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111339415B (en) | Click rate prediction method and device based on multi-interactive attention network | |
| Ho et al. | Predicting property prices with machine learning algorithms | |
| Han et al. | Semi-supervised active learning for sound classification in hybrid learning environments | |
| Chen et al. | Deep reinforcement learning in recommender systems: A survey and new perspectives | |
| Xing et al. | A hierarchical attention model for rating prediction by leveraging user and product reviews | |
| Xiang et al. | Answer selection in community question answering via attentive neural networks | |
| US11381651B2 (en) | Interpretable user modeling from unstructured user data | |
| CN109034203B (en) | Method, device, equipment and medium for training expression recommendation model and recommending expression | |
| CN112100375A (en) | Text information generation method and device, storage medium and equipment | |
| Liu et al. | CoFiGAN: Collaborative filtering by generative and discriminative training for one-class recommendation | |
| US20230196406A1 (en) | Siamese neural network model | |
| CN111209351A (en) | Object relation prediction method and device, object recommendation method and device, electronic equipment and medium | |
| CN112069404A (en) | Commodity information display method, device, equipment and storage medium | |
| Khan et al. | Comparative analysis on Facebook post interaction using DNN, ELM and LSTM | |
| CN117217284A (en) | Data processing method and device | |
| CN117216535A (en) | Training method, device, equipment and medium for recommended text generation model | |
| CN112949628B (en) | Track data enhancement and track identification method based on embedding-mixing | |
| CN114118526A (en) | Enterprise risk prediction method, device, equipment and storage medium | |
| Vielma et al. | Sentiment analysis with novel GRU based deep learning networks | |
| CN111445271A (en) | Model generation method, and prediction method, system, device and medium for cheating hotel | |
| CN110851600A (en) | Text data processing method and device based on deep learning | |
| CN112632275B (en) | Crowd clustering data processing method, device and equipment based on personal text information | |
| CN115186085A (en) | Reply content processing method and interaction method of media content interaction content | |
| CN114926210A (en) | Comment generation method, device, equipment and computer-readable storage medium | |
| CN111177493B (en) | Data processing method, device, server and storage 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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201211 |
|
| RJ01 | Rejection of invention patent application after publication |