CN111611496A - Product recommendation method and device - Google Patents
Product recommendation method and device Download PDFInfo
- Publication number
- CN111611496A CN111611496A CN202010274379.0A CN202010274379A CN111611496A CN 111611496 A CN111611496 A CN 111611496A CN 202010274379 A CN202010274379 A CN 202010274379A CN 111611496 A CN111611496 A CN 111611496A
- Authority
- CN
- China
- Prior art keywords
- user
- product
- users
- target
- behavior
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000011156 evaluation Methods 0.000 claims description 51
- 238000004364 calculation method Methods 0.000 claims description 20
- 238000012795 verification Methods 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 8
- 230000006399 behavior Effects 0.000 description 147
- 238000004422 calculation algorithm Methods 0.000 description 10
- 238000012549 training Methods 0.000 description 8
- 238000001914 filtration Methods 0.000 description 6
- 238000004891 communication Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000010365 information processing Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 238000011161 development Methods 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
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- 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]
- G06Q30/0631—Item recommendations
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- Data Mining & Analysis (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 application discloses a product recommendation method and a product recommendation device, and relates to the technical field of internet. The method comprises the following steps: determining a product tendency value of a user according to the historical behavior data of the user of the multi-dimensional sub-levels under different user behavior levels, the weights of the user behavior levels and the weights of the multi-dimensional sub-levels; determining the tendency value of the target user to the target product by acquiring the tendency values of a plurality of groups of users to different products; and when the tendency value meets a preset condition, sending the target product information corresponding to the tendency value to the target user. Through the technical scheme, products which are not concerned by the user but are interesting to the user can be mined and predicted, the purpose of recommending the products is achieved, the comprehensiveness of recommending the products is effectively improved, and the experience requirements of the user are met.
Description
Technical Field
The application relates to the technical field of internet, in particular to a product recommendation method and a product recommendation device.
Background
With the continuous development of the mobile internet, users prefer to acquire interesting service contents through platforms such as application programs providing business services, and the demand for personalized recommendation functions of the platforms such as the application programs is higher and higher. The personalized recommendation function of a platform such as an application program providing a business service in the prior art is generally to determine a product or a product attribute value interested by a user according to offline data of the user, so as to recommend the product interested by the user or a product matched with the product attribute value.
However, the prior art has the disadvantages that with the rapid increase of product types, a lot of products which are not concerned by the user but are interested by the user exist, so that the product which is interested by the user or the product attribute value is determined only according to the offline data of the user, and the product which is not concerned by the user but is interested by the user cannot be obtained, that is, the user does not have related browsing records, and the product which is not concerned by the user but is interested by the user cannot be determined through the offline data of the user, so that the comprehensiveness of product recommendation cannot be ensured, the experience requirements of the user cannot be accurately reflected, and the user experience is not good.
Disclosure of Invention
In view of this, the present application provides a product recommendation method and an apparatus thereof, and mainly aims to solve the technical problem that in the prior art, products that are not concerned by a user but are interesting to the user cannot be obtained according to offline data of the user, so that comprehensiveness of product recommendation is poor.
According to one aspect of the present application, there is provided a product recommendation method, the method comprising:
determining a product tendency value of a user according to the historical behavior data of the user of the multi-dimensional sub-levels under different user behavior levels, the weights of the user behavior levels and the weights of the multi-dimensional sub-levels;
determining the tendency value of the target user to the target product by acquiring the tendency values of a plurality of groups of users to different products;
and when the tendency value meets a preset condition, sending the target product information corresponding to the tendency value to the target user.
Preferably, the user behavior hierarchy comprises a user traffic behavior type and a user transaction behavior type; according to the multi-dimensional sub-hierarchy, the user flow behavior types comprise a browsing record dimension and a product collection record dimension, and the user transaction behavior types comprise a consumption record dimension and a product evaluation record dimension.
Preferably, the determining a product tendency value of a user according to the historical behavior data of the user at the multidimensional sublevel under different user behavior levels, the weights of the user behavior levels and the weights of the multidimensional sublevel specifically includes:
respectively determining each user behavior level weight and each multi-dimensional sub-level weight according to the user behavior level features and the multi-dimensional sub-level features;
according to the historical behavior data of the users of the multi-dimensional sub-levels under different user behavior levels, calculating the tendency values of the users to the products under different user behavior levels by using the multi-dimensional sub-level weights;
and calculating the tendency value of the user to the product by utilizing the user behavior hierarchy weight according to the tendency value of the user to the product under different user behavior hierarchies.
Preferably, the determining the tendency value of the target user to the target product by acquiring the tendency values of a plurality of groups of users to different products specifically includes:
calculating user weights of the plurality of groups of users relative to a target user;
and calculating the tendency value of the target user to the target product by utilizing the user weights of the plurality of groups of users relative to the target user according to the acquired tendency values of the plurality of groups of users to different products.
Preferably, the calculating the user weights of the plurality of groups of users relative to the target user specifically includes:
acquiring user attribute data of the plurality of groups of users;
and respectively setting the user weights of the plurality of groups of users relative to the target user according to the user attribute types of the user attribute data.
Preferably, the user attribute types include product evaluation and product purchase records, and the user weights of the multiple groups of users relative to the target user are respectively set according to the user attribute types of the user attribute data, specifically including:
according to the product evaluation information and the product consumption information in the user attribute data, respectively determining the similarity between a plurality of groups of users and a target user;
and respectively setting the user weights of the plurality of groups of users relative to the target user according to the similarity between the different users.
Preferably, the user attribute type further includes product evaluation time and product verification, and the similarity between the plurality of groups of users and the target user is respectively determined according to the product evaluation information and the product consumption information in the user attribute data, specifically including:
calculating similarity calculation weight according to product evaluation time information and product verification and cancellation information in the user attribute data;
and according to the product evaluation information and the product consumption information in the user attribute data, respectively determining the similarity between the plurality of groups of users and the target user by utilizing the similarity calculation weight.
According to still another aspect of the present application, there is provided a product recommendation apparatus including:
the hierarchy module is used for determining the tendency value of the user to the product according to the historical behavior data of the user of the multi-dimensional sub-hierarchy under different user behavior hierarchies, the weight of each user behavior hierarchy and the weight of each multi-dimensional sub-hierarchy;
the prediction module is used for determining the tendency value of the target user to the target product by acquiring the tendency values of a plurality of groups of users to different products;
and the recommending module is used for sending the product information corresponding to the tendency value to the target user when the tendency value meets a preset condition.
Preferably, the user behavior hierarchy comprises a user traffic behavior type and a user transaction behavior type; according to the multi-dimensional sub-hierarchy, the user flow behavior types comprise a browsing record dimension and a product collection record dimension, and the user transaction behavior types comprise a consumption record dimension and a product evaluation record dimension.
Preferably, the hierarchy module specifically includes:
the hierarchical weight unit is used for respectively determining each user behavior hierarchical weight and each multi-dimensional sub-hierarchical weight according to the user behavior hierarchical features and the multi-dimensional sub-hierarchical features;
the first tendency value unit is used for calculating tendency values of users to products under different user behavior levels by utilizing the multi-dimensional sub-level weights according to the historical user behavior data of the multi-dimensional sub-levels under different user behavior levels;
and the second tendency value unit is used for calculating the tendency value of the user to the product by utilizing the user behavior hierarchy weight according to the tendency value of the user to the product under different user behavior hierarchies.
Preferably, the prediction module specifically includes:
the first calculation unit is used for calculating the user weights of the plurality of groups of users relative to the target user;
and the second calculating unit is used for calculating the tendency value of the target user to the target product by utilizing the user weights of the plurality of groups of users relative to the target user according to the acquired tendency values of the plurality of groups of users to different products.
Preferably, the first computing unit specifically includes:
acquiring user attribute data of the plurality of groups of users;
and respectively setting the user weights of the plurality of groups of users relative to the target user according to the user attribute types of the user attribute data.
Preferably, the user attribute types include product evaluation and product purchase records, and the user weights of the multiple groups of users relative to the target user are respectively set according to the user attribute types of the user attribute data, specifically including:
according to the product evaluation information and the product consumption information in the user attribute data, respectively determining the similarity between a plurality of groups of users and a target user;
and respectively setting the user weights of the plurality of groups of users relative to the target user according to the similarity between the different users.
Preferably, the user attribute type further includes product evaluation time and product verification, and the similarity between the plurality of groups of users and the target user is respectively determined according to the product evaluation information and the product consumption information in the user attribute data, specifically including:
calculating similarity calculation weight according to product evaluation time information and product verification and cancellation information in the user attribute data;
and according to the product evaluation information and the product consumption information in the user attribute data, respectively determining the similarity between the plurality of groups of users and the target user by utilizing the similarity calculation weight.
According to yet another aspect of the present application, there is provided a storage medium having stored thereon a computer program which, when executed by a processor, implements the above-described product recommendation method.
According to yet another aspect of the present application, there is provided an apparatus comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, the processor implementing the product recommendation method when executing the program.
Compared with the prior art, the product recommendation method and the product recommendation device provided by the application have the advantages that the trend values of the users to the products are determined according to the historical behavior data of the users of the multi-dimensional sub-levels under different user behavior levels, the weights of the user behavior levels and the weights of the multi-dimensional sub-levels, the trend values of a plurality of groups of users to different products are determined according to the method for determining the trend values, so that the trend values of the target users to the target products are predicted, and when the trend values meet preset conditions, the target product information corresponding to the trend values is sent to the target users, so that the product recommendation to the target users is realized. Therefore, the tendency values of a plurality of groups of users to different products are calculated in a collaborative filtering mode by utilizing the historical behavior data of the users, and the tendency values of the target users to the target products which are not concerned (not scored) are predicted, so that the products which are not concerned by the users but are interested by the users are mined and predicted, the purpose of recommending the products is achieved, the comprehensiveness of recommending the products is effectively improved, and the experience requirements of the users are met.
The foregoing description is only an overview of the technical solutions of the present application, and the present application can be implemented according to the content of the description in order to make the technical means of the present application more clearly understood, and the following detailed description of the present application is given in order to make the above and other objects, features, and advantages of the present application more clearly understandable.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the application and together with the description serve to explain the application and not to limit the application. In the drawings:
FIG. 1 is a flow chart illustrating a method for recommending a product according to an embodiment of the present application;
FIG. 2 is a flow chart illustrating another product recommendation method provided by an embodiment of the present application;
FIG. 3 is a schematic structural diagram illustrating a product recommendation device according to an embodiment of the present application;
fig. 4 shows a schematic structural diagram of another product recommendation device provided in an embodiment of the present application.
Detailed Description
The present application will be described in detail below with reference to the accompanying drawings in conjunction with embodiments. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
For solving the problems existing in the prior art, the embodiment provides a product recommendation method, which calculates tendency values of a plurality of groups of users to different products by using a collaborative filtering manner, and then predicts the tendency values of target users to target products that are not concerned, so as to mine and predict products that are not concerned but are interested by the users, and effectively improve the comprehensiveness of product recommendation, so as to meet the experience requirements of the users on personalized recommendation, as shown in fig. 1, the method includes:
In this embodiment, the user historical behavior data is divided into two user behavior levels according to the user behavior type, which are a user traffic behavior type and a user transaction behavior type, and different multidimensional sub-levels are set according to different user behavior levels, for example, the user traffic behavior type includes a browsing record dimension and a product collection record dimension, and the user transaction behavior type includes a consumption record dimension and a product evaluation record dimension. The user behavior hierarchy includes a plurality of user behavior sub-levels, wherein the user behavior sub-levels include a plurality of user behavior sub-levels, and the user behavior sub-levels include a plurality of user behavior sub-levels.
Correspondingly, user historical behavior data of a certain user are obtained, corresponding user historical behavior data are obtained according to different user behavior levels and different multi-dimensional sub-levels of the user historical behavior data, data characteristics of a certain product under different user behavior levels and different multi-dimensional sub-levels are obtained according to the obtained user historical behavior data, the data characteristics can be characteristic vectors and characteristic values according to requirements of an actual application scene, and the data characteristics are not specifically limited.
Further, according to data characteristics of different user behavior levels and different multi-dimensional sub-level levels, the multi-dimensional sub-level weights are used for carrying out level tendency value calculation to obtain multi-dimensional sub-level scores of a certain product by a certain user, so that a user behavior level tendency value is obtained, further according to the user behavior level tendency value, the user tendency value to the product is obtained by the user behavior level weights, and the accuracy of the calculated user tendency value to the product is higher.
And 102, determining the tendency value of the target user to the target product by acquiring the tendency values of the plurality of groups of users to different products.
In this embodiment, the method in step 101 is utilized to obtain user historical behavior data of a certain user, which includes transaction records of multiple products, and calculate tendency values of the certain user to the multiple products respectively according to the user historical behavior data of the multiple products, and similarly, obtain multiple sets of user historical behavior data, calculate tendency values of the multiple sets of users to the multiple products, and calculate tendency values of the target user to the target product by utilizing a Weighted Slope One (WSO: Weighted Slope One) algorithm, where the multiple products of the multiple sets of users include the target product. In the embodiment, the user weight factor is introduced into the WSO algorithm, so that the accuracy of the calculated trend value of the target user to the target product is higher, and the user weight factor is determined based on the similarity between multiple groups of users and the target user.
And 103, when the tendency value meets a preset condition, sending the target product information corresponding to the tendency value to the target user.
In this embodiment, when the tendency value of the target product by the target user exceeds the preset tendency value, it indicates that the target product is a product in which the target user is interested, obtains link information of the target product and sends the link information to the target user, and according to the requirement of the actual application scenario, the link information is presented in a link manner of one or more of a product image, a product video and a product text introduction, and the link information is not specifically limited here.
Compared with the prior art, the product recommendation method provided by the embodiment can determine the tendency values of the users to the products according to the historical behavior data of the users of the multi-dimensional sub-levels under different user behavior levels, the weights of the user behavior levels and the weights of the multi-dimensional sub-levels, and determine the tendency values of a plurality of groups of users to different products according to the method for determining the tendency values, so that the tendency values of the target users to the target products are predicted, and when the tendency values meet preset conditions, target product information corresponding to the tendency values is sent to the target users, so that the product recommendation to the target users is realized. Therefore, the tendency values of a plurality of groups of users to different products are calculated in a collaborative filtering mode by utilizing the historical behavior data of the users, and then the tendency values of target products which are not concerned with intersection, namely are not concerned with, are predicted for the target users, so that products which are not concerned with but are interested by the users are mined and predicted, the purpose of product recommendation is achieved, the comprehensiveness of product recommendation is effectively improved, and the experience requirements of the users are met.
Further, as a refinement and an extension of the specific implementation of the above embodiment, in order to fully explain the process in the present embodiment, another product recommendation method is provided, as shown in fig. 2, and the method includes:
Further, for illustrating the specific implementation process of step 201, as an optional way, the user behavior hierarchy includes a user traffic behavior type and a user transaction behavior type; according to the multi-dimensional sub-hierarchy, the user flow behavior types comprise a browsing record dimension and a product collection record dimension, and the user transaction behavior types comprise a consumption record dimension and a product evaluation record dimension.
In specific implementation, by acquiring a plurality of groups of user historical behavior data samples of user browsing record dimensionality and product collection record dimensionality of a certain class of products and preset values of user flow behaviors of the class of products, weights of browsing record dimensionality sub-levels and product collection record dimensionality sub-levels are obtained by utilizing a parameter tuning algorithm for training; acquiring a plurality of groups of user historical behavior data samples of consumption record dimensionality and product evaluation record dimensionality of a certain class of products by users, and acquiring weights of consumption record dimensionality sublevels and product evaluation record dimensionality sublevels of the class of products by utilizing a parameter tuning algorithm for training according to preset values of user transaction behaviors of the class of products; and training by utilizing a parameter tuning algorithm based on the obtained multiple groups of user flow behavior types of the users for certain products and the user historical behavior data samples of the multi-dimensional sub-levels under the user transaction behavior types and the preset tendency values of the products of the class to obtain the weights of the user flow behavior types and the user transaction behavior types of the products of the class by the users.
The objective function for the browsing record dimension sublevel and the product collection record dimension sublevel weight is as follows: fw＝∑kμkfk；
Wherein, FwA predetermined tendency value, f, for the type of user traffic behavior1Score corresponding to a sample of historical behavior data of the user for a browsing record dimension, f2Obtaining the sub-level weight mu of the dimension of the browsing record by training a parameter tuning algorithm for the score corresponding to the user historical behavior data sample of the dimension of the product collection record, k ∈ 21And product collection record dimension sub-level weight mu2。
The objective function for the consumption record dimension sub-level and the product evaluation record dimension sub-level weights is: t isr＝∑x xtrx；
Wherein, TrA predetermined tendency value, tr, for the type of transaction behaviour of the user1Rating, tr, corresponding to a sample of user historical behavior data for a consumption record dimension2Obtaining the sub-level weight of the dimension of the consumption record by training a parameter tuning algorithm for the score corresponding to the user historical behavior data sample of the dimension of the product evaluation record, x ∈ 21And product evaluation record dimension sublevel weight2。
The objective function for the user traffic behavior type hierarchy and the user transaction behavior type hierarchy weight is:
wherein L is a preset tendency value of a user to a certain class of products, and the user flow behavior type level weight and the user transaction behavior type level weight are obtained through parameter tuning algorithm training
Presetting the hierarchical weight of the transaction behavior type of the user according to the requirements of the actual application scene
Is greater than the hierarchical weight of the user flow behavior type, and when the hierarchical weight of the user flow behavior type obtained by the training of the parameter tuning algorithm is not greater than the hierarchical weight of the user transaction behavior type
And (3) carrying out hierarchical weight on the user traffic behavior type and the user transaction behavior type
Performing re-tuning until the user transaction behaviorType level weighting
Greater than the user traffic behavior type level weight.
And 202, calculating the tendency values of the users to the products under different user behavior levels by utilizing the multi-dimensional sub-level weights according to the historical user behavior data of the multi-dimensional sub-levels under different user behavior levels.
In specific implementation, taking the user flow behavior type including a browsing record dimension and a product collection record dimension as an example, the user historical behavior data of the browsing record dimension and the product collection record dimension are respectively obtained, that is, browsing record data of the user u on the product j is obtained, a data feature V (u, j) of the browsing times of the user u on the product j is determined, collecting record data of the user u on the product j is obtained, and a data feature C (u, j) of whether the user u performs a collecting operation on the product j is determined. Further, the data characteristics V (u, j) of the browsing times of the product j by the user u and the data characteristics C (u, j) of whether the user u collects the product j are respectively normalized to obtain scores of the multidimensional sub-levels V (u, j) and C (u, j), and the value range is between [0 and 1 ].
Similarly, for a consumption record dimension and a product evaluation record dimension included in the user transaction behavior type, user historical behavior data of the consumption record dimension and the product evaluation record dimension are respectively obtained, namely consumption record data of the user u on the product j are obtained, data characteristics P (u, j) of consumption information of the user u on the product j are determined, rating record data of the user u on the product j are obtained, whether the user u evaluates the data characteristics S (u, j) of the product j is determined, after normalization processing is also carried out, scores of multi-dimensional sub-levels P (u, j) and S (u, j) are obtained, and the value range is between [0 and 1 ]. According to the requirements of the actual application scene, the data characteristics P (u, j) of the consumption information include consumption amount, consumption times and the like, the corresponding characteristic vector is [ consumption amount, consumption times ], and the evaluation grade S (u, j) can be star-grade evaluation of the product, and the contents of the consumption information and the evaluation grade are not specifically limited here.
Correspondingly, for the user traffic behavior type, the grade f of the V (u, j) is obtained1Score f of C (u, j)2And the weight mu corresponding to the trained V (u, j)1Weight μ corresponding to C (u, j)2Calculating the tendency value F of the user u to the product j under the user flow behavior typew(ii) a Similarly, for the user transaction behavior type, the score tr of P (u, j) is obtainediScore tr of S (u, j)2And the weight corresponding to the trained P (u, j)1Weight corresponding to S (u, j)2Calculating the tendency value T of the user u to the product j under the transaction behavior type of the userr。
And 203, calculating the tendency value of the user to the product by utilizing the user behavior hierarchy weight according to the tendency value of the user to the product under different user behavior hierarchies.
In specific implementation, according to the proportion of the tendency value of the user traffic behavior type in the user behavior type and the proportion of the tendency value of the user transaction behavior type in the user behavior type, the training is utilized to obtain the hierarchical weight of the user traffic behavior type and the hierarchical weight of the user transaction behavior type
And calculating the tendency value of the user to the product.
And 204, calculating the user weights of the multiple groups of users relative to the target user.
Further, for explaining a specific implementation process of step 204, as an optional manner, step 204 may specifically include:
step 2041, obtain the user attribute data of the multiple groups of users.
Step 2042, respectively setting the user weights of the multiple groups of users relative to the target user according to the user attribute types of the user attribute data.
Further, for explaining the specific implementation process of step 2042, as an alternative, the user attribute types include product evaluation and product purchase records, and step 2042 may specifically include: according to the product evaluation information and the product consumption information in the user attribute data, respectively determining the similarity between a plurality of groups of users and a target user; and respectively setting the user weights of the plurality of groups of users relative to the target user according to the similarity between the different users.
In specific implementation, product evaluation information and product purchase record information in the user attribute data of the multiple groups of users are obtained, according to the requirements of an actual application scene, the product evaluation information is the value of the user as a product, namely the comment content of the user on the product is obtained, and the value of the comment content is obtained by identifying and quantifying the comment content; the product purchase record information is the offline consumption expense amount of the product by the user, normalization processing is respectively carried out, the value range is between [0 and 1], and the content of the product evaluation information and the product purchase record information is not specifically limited.
Correspondingly, according to the product evaluation information and the product consumption information in the user attribute data, the similarity between the plurality of groups of users and the target user is respectively determined, and the calculation formula is specifically as follows:
where Sim (u, v) is the similarity between user u and target user v, Rv,jA value of credit, R, to product j for target user vu,jFor the value of the user u's score for product j,
the average value of the scores of the products that the target user v has scored,
the average value of the scores of the products scored by the user u; pv,jAmount of offline payment, P, for target user v for product ju,jThe user u pays the amount of the expense for the offline payment of the product j,
lines over a period of time for a target user vThe average value of the amount of the lower expense verification and cancellation,
average value of offline payment amounts for user u over a period of time, Iu,vA collection of products that have been jointly scored for user u and target user v.
Further, the user attribute type further includes product evaluation time and product verification, and the determining of the similarity between different users according to the product evaluation information and the product consumption information in the user attribute data specifically includes:
calculating similarity calculation weight according to product evaluation time information and product verification and cancellation information in the user attribute data; and according to the product evaluation information and the product consumption information in the user attribute data, respectively determining the similarity between the plurality of groups of users and the target user by utilizing the similarity calculation weight.
In specific implementation, the product marketing information in the user attribute data is a spatial distance between position information when a user purchases a product and position information when the user markets the product, and when the user attribute type further includes product evaluation time and product marketing, in order to improve accuracy of similarity between users and make the similarity more consistent with scene characteristics of local life, similarity calculation weights are set for the similarity calculation formulas, that is, the improved calculation formulas for respectively determining similarities between a plurality of groups of users and target users are:
wherein α is the time adjustment coefficient, tu,jTime for user u to evaluate product after offline expense verification of product j, tv,jTime for evaluating the product after the offline expense verification of the product j for the target user v, β space adjustment coefficient, du,jAnd the space distance between the positioning address when the product j is purchased for the user u and the positioning address of the product j which is consumed and sold under the line of the user u.
Further, a user similarity calculation formula is utilized to calculate a similarity value between the user u and the target user v, when a tendency value of the target user v to the target product is calculated, according to the similarity value between the user u and the target user v, a user weight of the user u relative to the target user v, namely 1-Sim (u, v), is given, and user weights of a plurality of groups of users relative to the target user v are calculated in sequence.
And step 205, calculating the tendency values of the target users to the target products by using the user weights of the plurality of groups of users relative to the target users according to the acquired tendency values of the plurality of groups of users to different products.
In specific implementation, the WSO algorithm is improved by introducing the user weight between the user and the target user, namely, the similarity between the user and the target user is higher (namely, the smaller the Sim (u, v) value is), the larger the user weight value for prediction is, so as to improve the calculation accuracy of the target user on the tendency value of the target product.
The specific calculation formula is as follows:
wherein,
predicted tendency value, S, for target user v to target product xuProduct collections with purchasing behavior for groups of users, rukThe tendency values of different users (1-u) to different products (1-k) are obtained by the calculation method of the step 202, dxkThe difference of the scores of the target product x and different products, cxkThe number of users who buy the target product x and the product k at the same time, or the number of operations based on the target product x and the product k.
And step 206, when the tendency value meets a preset condition, sending the product information corresponding to the tendency value to the target user.
In specific implementation, if the calculated predicted tendency value of the target user v on the target product x exceeds the preset tendency value, the target product x is a product which the target user v does not care but is interested in, and if the calculated predicted tendency value of the target user v on the target product x does not exceed the preset tendency value, the target product x does not belong to a product which the target user v is interested in, and is not recommended to the target user.
Based on the requirements of the actual application scene, based on the method for calculating the tendency values, the tendency values of the target users to the target products are calculated, the target products with the predicted tendency values exceeding the preset tendency values are sorted according to the predicted tendency values, and a product list corresponding to the sorting result is sent to the target users, so that the target users can be intelligently recommended, and can also be recommended to the users after being mixed with traditional recommended products, wherein the combined recommendation mode is not specifically limited.
By applying the method provided by the embodiment, the tendency value of the user to the product is determined according to the user historical behavior data of the multi-dimensional sub-levels under different user behavior levels, the weight of each user behavior level and the weight of each multi-dimensional sub-level, and the tendency values of a plurality of groups of users to different products are determined according to the method for determining the tendency values, so that the tendency value of the target user to the target product is predicted, and when the tendency values meet the preset condition, the target product information corresponding to the tendency values is sent to the target user, so that the product recommendation to the target user is realized. Therefore, the tendency values of a plurality of groups of users to different products are calculated by utilizing the historical behavior data of the users in a collaborative filtering mode, and then the tendency values of the target users to the target products which are not concerned by the users are predicted, so that the products which are not concerned by the users but are interested by the users are mined and predicted, the purpose of recommending the products is achieved, the comprehensiveness of recommending the products is effectively improved, and the experience requirements of the users are met.
Further, as a specific implementation of the method shown in fig. 1 and fig. 2, an embodiment of the present application provides a product recommendation device, as shown in fig. 3, the device includes: a hierarchy module 31, a prediction module 32, a recommendation module 33.
The hierarchy module 31 is configured to determine a trend value of the user to the product according to the historical behavior data of the user at the multi-dimensional sub-hierarchy under different user behavior hierarchies, and the weight of each user behavior hierarchy and the weight of each multi-dimensional sub-hierarchy.
And the prediction module 32 is used for determining the tendency value of the target user to the target product by acquiring the tendency values of a plurality of groups of users to different products.
And the recommending module 33 is configured to send the product information corresponding to the tendency value to the target user when the tendency value meets a preset condition.
In a specific application scenario, as shown in fig. 4, the user behavior hierarchy includes a user traffic behavior type and a user transaction behavior type; according to the multi-dimensional sub-hierarchy, the user flow behavior types comprise a browsing record dimension and a product collection record dimension, and the user transaction behavior types comprise a consumption record dimension and a product evaluation record dimension.
In a specific application scenario, the hierarchy module 31 specifically includes: a hierarchy weight unit 311, a first tendency value unit 312, and a second tendency value unit 313.
The hierarchical weight unit 311 is specifically configured to determine each user behavior hierarchical weight and each multidimensional sublevel weight according to the user behavior hierarchical feature and the multidimensional sublevel feature.
The first tendency value unit 312 is specifically configured to calculate, according to the historical behavior data of the users at the multidimensional sub-levels under different user behavior levels, tendency values of the users to the product under different user behavior levels by using the multidimensional sub-level weights.
The second tendency value unit 313 is specifically configured to calculate, according to tendency values of users to products under different user behavior hierarchies, tendency values of users to products by using the user behavior hierarchy weights.
In a specific application scenario, the prediction module 32 specifically includes: a first calculating unit 321 and a second calculating unit 322.
The first calculating unit 321 is specifically configured to calculate user weights of the multiple groups of users relative to the target user.
The second calculating unit 322 is specifically configured to calculate, according to the obtained tendency values of the multiple groups of users to different products, a tendency value of the target user to the target product by using the user weights of the multiple groups of users relative to the target user.
In a specific application scenario, the first calculating unit 321 specifically includes: acquiring user attribute data of the plurality of groups of users; and respectively setting the user weights of the plurality of groups of users relative to the target user according to the user attribute types of the user attribute data.
In a specific application scenario, the user attribute types include product evaluation and product purchase records, and the user weights of the multiple groups of users relative to the target user are respectively set according to the user attribute types of the user attribute data, specifically including: according to the product evaluation information and the product consumption information in the user attribute data, respectively determining the similarity between a plurality of groups of users and a target user; and respectively setting the user weights of the plurality of groups of users relative to the target user according to the similarity between the different users.
In a specific application scenario, the user attribute type further includes product evaluation time and product verification, and the similarity between a plurality of groups of users and a target user is respectively determined according to product evaluation information and product consumption information in the user attribute data, specifically including: calculating similarity calculation weight according to product evaluation time information and product verification and cancellation information in the user attribute data; and according to the product evaluation information and the product consumption information in the user attribute data, respectively determining the similarity between the plurality of groups of users and the target user by utilizing the similarity calculation weight.
It should be noted that other corresponding descriptions of the functional modules and functional units related to the product recommendation device provided in this embodiment may refer to the corresponding descriptions in fig. 1 and fig. 2, and are not described herein again.
By applying the product recommending device, the tendency value of the user to the product is determined according to the user historical behavior data of the multi-dimensional sub-levels under different user behavior levels, the weight of each user behavior level and the weight of each multi-dimensional sub-level, and the tendency values of a plurality of groups of users to different products are determined according to the method for determining the tendency values, so that the tendency value of the target user to the target product is predicted, and when the tendency values meet preset conditions, target product information corresponding to the tendency values is sent to the target user, so that the product recommendation to the target user is realized. Therefore, the tendency values of a plurality of groups of users to different products are calculated in a collaborative filtering mode by utilizing the historical behavior data of the users, and then the tendency values of target products which are not concerned with intersection, namely are not concerned with, are predicted for the target users, so that products which are not concerned with but are interested by the users are mined and predicted, the purpose of product recommendation is achieved, the comprehensiveness of product recommendation is effectively improved, and the experience requirements of the users are met.
Based on the above methods shown in fig. 1 and fig. 2, correspondingly, the present application further provides a storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the method for recommending a product shown in fig. 1 and fig. 2 is implemented.
Based on such understanding, the technical solution of the present application may be embodied in the form of a software product, which may be stored in a non-volatile storage medium (which may be a CD-ROM, a usb disk, a removable hard disk, or the like), and includes several instructions for enabling a computer device (which may be a personal computer, a server, or a network device, or the like) to execute the product recommendation method according to the various implementation scenarios of the present application.
Based on the method shown in fig. 1 and fig. 2 and the virtual device embodiment shown in fig. 3 and fig. 4, in order to achieve the above object, an embodiment of the present application further provides a terminal device, which may specifically be a personal computer, a tablet computer, a smart phone, a smart watch, a POS device, or other network devices, and the terminal device includes a storage medium and a processor; a storage medium for storing a computer program; a processor for executing a computer program to implement the product recommendation method as described above with reference to fig. 1 and 2.
Optionally, the above entity devices may further include a user interface, a network interface, a camera, a Radio Frequency (RF) circuit, a sensor, an audio circuit, a WI-FI module, and the like. The user interface may include a Display screen (Display), an input unit such as a keypad (Keyboard), etc., and the optional user interface may also include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface), etc.
Those skilled in the art will appreciate that the physical device structure of a terminal device provided in this embodiment is not limited to the above physical device, and may include more or less components, or combine some components, or arrange different components.
The storage medium may further include an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the two physical devices described above, supporting the operation of the information processing program as well as other software and/or programs. The network communication module is used for realizing communication among components in the storage medium and communication with other hardware and software in the information processing entity device.
Through the above description of the embodiments, those skilled in the art will clearly understand that the present application can be implemented by software plus a necessary general hardware platform, and can also be implemented by hardware. Through the application of the technical scheme, compared with the prior art, the method and the device have the advantages that the historical behavior data of the users are utilized, the tendency values of the multiple groups of users to different products are calculated in a collaborative filtering mode, and then the tendency values of target products which are not concerned with intersection, namely not concerned with the target users are predicted, so that the products which are not concerned with the users but are interested by the users are mined and predicted, the purpose of product recommendation is achieved, the comprehensiveness of product recommendation is effectively improved, and the experience requirements of the users are met.
Those skilled in the art will appreciate that the figures are merely schematic representations of one preferred implementation scenario and that the blocks or flow diagrams in the figures are not necessarily required to practice the present application. Those skilled in the art will appreciate that the modules in the devices in the implementation scenario may be distributed in the devices in the implementation scenario according to the description of the implementation scenario, or may be located in one or more devices different from the present implementation scenario with corresponding changes. The modules of the implementation scenario may be combined into one module, or may be further split into a plurality of sub-modules.
The above application serial numbers are for description purposes only and do not represent the superiority or inferiority of the implementation scenarios. The above disclosure is only a few specific implementation scenarios of the present application, but the present application is not limited thereto, and any variations that can be made by those skilled in the art are intended to fall within the scope of the present application.
Claims (10)
1. A method for recommending products, comprising:
determining a product tendency value of a user according to the historical behavior data of the user of the multi-dimensional sub-levels under different user behavior levels, the weights of the user behavior levels and the weights of the multi-dimensional sub-levels;
determining the tendency value of the target user to the target product by acquiring the tendency values of a plurality of groups of users to different products;
and when the tendency value meets a preset condition, sending the target product information corresponding to the tendency value to the target user.
2. The method of claim 1, wherein the user behavior hierarchy includes a user traffic behavior type, a user transaction behavior type; according to the multi-dimensional sub-hierarchy, the user flow behavior types comprise a browsing record dimension and a product collection record dimension, and the user transaction behavior types comprise a consumption record dimension and a product evaluation record dimension.
3. The method according to claim 1 or 2, wherein the determining the trend value of the user to the product according to the historical behavior data of the user at the multi-dimensional sub-level under different user behavior levels, the weights of the user behavior levels and the weights of the multi-dimensional sub-levels specifically comprises:
respectively determining each user behavior level weight and each multi-dimensional sub-level weight according to the user behavior level features and the multi-dimensional sub-level features;
according to the historical behavior data of the users of the multi-dimensional sub-levels under different user behavior levels, calculating the tendency values of the users to the products under different user behavior levels by using the multi-dimensional sub-level weights;
and calculating the tendency value of the user to the product by utilizing the user behavior hierarchy weight according to the tendency value of the user to the product under different user behavior hierarchies.
4. The method according to claim 1 or 2, wherein the determining the tendency value of the target user for the target product by obtaining the tendency values of a plurality of groups of users for different products specifically comprises:
calculating user weights of the plurality of groups of users relative to a target user;
and calculating the tendency value of the target user to the target product by utilizing the user weights of the plurality of groups of users relative to the target user according to the acquired tendency values of the plurality of groups of users to different products.
5. The method according to claim 4, wherein the calculating the user weights of the plurality of groups of users relative to the target user specifically comprises:
acquiring user attribute data of the plurality of groups of users;
and respectively setting the user weights of the plurality of groups of users relative to the target user according to the user attribute types of the user attribute data.
6. The method according to claim 5, wherein the user attribute types include product evaluation and product purchase records, and the setting of the user weights of the plurality of groups of users relative to the target user according to the user attribute types of the user attribute data respectively includes:
according to the product evaluation information and the product consumption information in the user attribute data, respectively determining the similarity between a plurality of groups of users and a target user;
and respectively setting the user weights of the plurality of groups of users relative to the target user according to the similarity between the different users.
7. The method according to claim 6, wherein the user attribute types further include product evaluation time and product verification, and the determining the similarity between the plurality of groups of users and the target user according to the product evaluation information and the product consumption information in the user attribute data specifically includes:
calculating similarity calculation weight according to product evaluation time information and product verification and cancellation information in the user attribute data;
and according to the product evaluation information and the product consumption information in the user attribute data, respectively determining the similarity between the plurality of groups of users and the target user by utilizing the similarity calculation weight.
8. A product recommendation device, comprising:
the hierarchy module is used for determining the tendency value of the user to the product according to the historical behavior data of the user of the multi-dimensional sub-hierarchy under different user behavior hierarchies, the weight of each user behavior hierarchy and the weight of each multi-dimensional sub-hierarchy;
the prediction module is used for determining the tendency value of the target user to the target product by acquiring the tendency values of a plurality of groups of users to different products;
and the recommending module is used for sending the product information corresponding to the tendency value to the target user when the tendency value meets a preset condition.
9. A storage medium on which a computer program is stored, the program, when executed by a processor, implementing the product recommendation method of any one of claims 1 to 7.
10. An apparatus comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor implements the product recommendation method of any one of claims 1 to 7 when executing the program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010274379.0A CN111611496A (en) | 2020-04-09 | 2020-04-09 | Product recommendation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010274379.0A CN111611496A (en) | 2020-04-09 | 2020-04-09 | Product recommendation method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111611496A true CN111611496A (en) | 2020-09-01 |
Family
ID=72199357
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010274379.0A Pending CN111611496A (en) | 2020-04-09 | 2020-04-09 | Product recommendation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111611496A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112579902A (en) * | 2020-12-24 | 2021-03-30 | 第四范式（北京）技术有限公司 | Behavior data management method and device supporting multiple intelligent application scenes |
| CN112598464A (en) * | 2020-12-22 | 2021-04-02 | 浙江敦奴联合实业股份有限公司 | Garment customization scheme recommendation method and device |
| CN112783984A (en) * | 2021-02-10 | 2021-05-11 | 中国人民银行数字货币研究所 | Information display method and device |
| CN113592588A (en) * | 2021-07-25 | 2021-11-02 | 北京慧橙信息科技有限公司 | E-commerce platform commodity recommendation system and method based on big data collaborative filtering technology |
| CN113807905A (en) * | 2020-11-05 | 2021-12-17 | 北京沃东天骏信息技术有限公司 | Article recommendation method and device, computer storage medium and electronic equipment |
| CN114880569A (en) * | 2022-05-18 | 2022-08-09 | 中国第一汽车股份有限公司 | Recommendation control method and device for vehicle, electronic equipment, system and storage medium |
| TWI827029B (en) * | 2022-04-29 | 2023-12-21 | 台灣伽瑪移動數位股份有限公司 | Method for recommending commodities and the related electronic device |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102254028A (en) * | 2011-07-22 | 2011-11-23 | 青岛理工大学 | Personalized commodity recommendation method and system integrating attributes and structural similarity |
| CN106779825A (en) * | 2016-12-02 | 2017-05-31 | 乐视控股（北京）有限公司 | A kind of item recommendation method, device and electronic equipment |
| CN106897911A (en) * | 2017-01-10 | 2017-06-27 | 南京邮电大学 | A kind of self adaptation personalized recommendation method based on user and article |
| US20170206581A1 (en) * | 2016-01-15 | 2017-07-20 | Target Brands, Inc. | Product vector for product recommendation |
| CN107295107A (en) * | 2017-08-01 | 2017-10-24 | 深圳天珑无线科技有限公司 | Recommendation method, recommendation apparatus and mobile terminal |
| CN110458637A (en) * | 2019-06-19 | 2019-11-15 | 中国平安财产保险股份有限公司 | Product method for pushing and its relevant device neural network based |
-
2020
- 2020-04-09 CN CN202010274379.0A patent/CN111611496A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102254028A (en) * | 2011-07-22 | 2011-11-23 | 青岛理工大学 | Personalized commodity recommendation method and system integrating attributes and structural similarity |
| US20170206581A1 (en) * | 2016-01-15 | 2017-07-20 | Target Brands, Inc. | Product vector for product recommendation |
| CN106779825A (en) * | 2016-12-02 | 2017-05-31 | 乐视控股（北京）有限公司 | A kind of item recommendation method, device and electronic equipment |
| CN106897911A (en) * | 2017-01-10 | 2017-06-27 | 南京邮电大学 | A kind of self adaptation personalized recommendation method based on user and article |
| CN107295107A (en) * | 2017-08-01 | 2017-10-24 | 深圳天珑无线科技有限公司 | Recommendation method, recommendation apparatus and mobile terminal |
| CN110458637A (en) * | 2019-06-19 | 2019-11-15 | 中国平安财产保险股份有限公司 | Product method for pushing and its relevant device neural network based |
Non-Patent Citations (1)
| Title |
|---|
| 刘林静;楼文高;冯国珍;: "基于用户相似性的加权Slope One算法" * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113807905A (en) * | 2020-11-05 | 2021-12-17 | 北京沃东天骏信息技术有限公司 | Article recommendation method and device, computer storage medium and electronic equipment |
| CN112598464A (en) * | 2020-12-22 | 2021-04-02 | 浙江敦奴联合实业股份有限公司 | Garment customization scheme recommendation method and device |
| CN112579902A (en) * | 2020-12-24 | 2021-03-30 | 第四范式（北京）技术有限公司 | Behavior data management method and device supporting multiple intelligent application scenes |
| CN112783984A (en) * | 2021-02-10 | 2021-05-11 | 中国人民银行数字货币研究所 | Information display method and device |
| CN113592588A (en) * | 2021-07-25 | 2021-11-02 | 北京慧橙信息科技有限公司 | E-commerce platform commodity recommendation system and method based on big data collaborative filtering technology |
| CN113592588B (en) * | 2021-07-25 | 2023-10-03 | 深圳市瀚力科技有限公司 | E-commerce platform commodity recommendation system and method based on big data collaborative filtering technology |
| TWI827029B (en) * | 2022-04-29 | 2023-12-21 | 台灣伽瑪移動數位股份有限公司 | Method for recommending commodities and the related electronic device |
| CN114880569A (en) * | 2022-05-18 | 2022-08-09 | 中国第一汽车股份有限公司 | Recommendation control method and device for vehicle, electronic equipment, system and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111611496A (en) | Product recommendation method and device | |
| CN108665355B (en) | Financial product recommendation method, apparatus, device and computer storage medium | |
| CN108648074B (en) | Loan assessment method, device and equipment based on support vector machine | |
| CN103678672B (en) | Method for recommending information | |
| KR20200123015A (en) | Information recommendation method, apparatus, device and medium | |
| CN108694647B (en) | Method and device for mining merchant recommendation reason and electronic equipment | |
| US20200294111A1 (en) | Determining target user group | |
| CN108805598B (en) | Similarity information determination method, server and computer-readable storage medium | |
| CN111626767B (en) | Resource data issuing method, device and equipment | |
| CN109299356B (en) | Activity recommendation method and device based on big data, electronic equipment and storage medium | |
| CN111061979B (en) | User tag pushing method and device, electronic equipment and medium | |
| CN102934113A (en) | Information provision system, information provision method, information provision device, program, and information recording medium | |
| CN110909222A (en) | User portrait establishing method, device, medium and electronic equipment based on clustering | |
| CN109902713A (en) | Building recommended method, equipment, storage medium and device based on data analysis | |
| WO2023000491A1 (en) | Application recommendation method, apparatus and device, and computer-readable storage medium | |
| CN112449002B (en) | Method, device and equipment for pushing object to be pushed and storage medium | |
| CN108932658B (en) | Data processing method, device and computer readable storage medium | |
| CN115757952A (en) | Content information recommendation method, device, equipment and storage medium | |
| CN110348947A (en) | Object recommendation method and device | |
| CN106447364A (en) | Information recommending method and information recommending device | |
| CN111754287A (en) | Article screening method, apparatus, device and storage medium | |
| CN115271931A (en) | Credit card product recommendation method and device, electronic equipment and medium | |
| CN112036987B (en) | Method and device for determining recommended commodity | |
| CN112015970A (en) | Product recommendation method, related equipment and computer storage medium | |
| CN110490749A (en) | A kind of price fixing method and device |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200901 |