Recommendation pushing method based on information and neural network
By obtaining basic information and daily records with the user's consent, dividing user groups, and using neural networks to analyze interests, the lag problem of existing recommendation systems is solved, and the precise locking of user interests and improving recommendation effects are achieved.
Patent Information
- Application Number
- CN202411871374.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-18
- Publication Date
- 2025-03-28
AI Technical Summary
The existing recommendation system has lag and cannot predict user interest points in advance, resulting in poor recommendation results.
By obtaining basic information and daily records with the user's consent, dividing the user groups and making preliminary recommendations at the appropriate time. Use neural networks to analyze group interests, obtain stage data, analyze the degree of interest in information in advance, and accurately plan the aspects of users' interest and product push.
It realizes accurate locking of user interests, improves the effect and accuracy of information recommendations, and can continuously update the recommended content according to different stages of the user.
Smart Images
Figure CN119722244A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of information push, and more specifically, to a recommendation push method based on information and neural network. Background Art
[0002] With the explosive growth of Internet information, it is becoming increasingly difficult for users to obtain personalized information. As an indispensable tool for information filtering, the recommendation system can quickly and accurately find content that users may be interested in from massive data. It is no exaggeration to say that recommendation systems are now ubiquitous on the Internet and play a vital role in many user-oriented online services.
[0003] Nowadays, researchers have proposed a large number of recommendation algorithms. Among these existing recommendation technologies, collaborative filtering is the most outstanding method and has been widely used in the industry. However, the above recommendation methods all have a certain lag, cannot predict the user's interests in advance, and cannot recommend relevant products in a targeted manner, which makes the final recommendation effect poor despite the complexity of the algorithm. Summary of the invention
[0004] The purpose of the present invention is to provide a recommendation push method based on information and neural networks, aiming to solve the problem of having a certain lag, being unable to predict the user's interests in advance, and being unable to recommend relevant products in a targeted manner, resulting in a poor final recommendation effect despite the complexity of the algorithm.
[0005] To achieve the above purpose, the technical solution adopted by the present invention is: to provide a recommendation push method based on information and neural network, including: Obtaining basic information of the user with the user's consent, and dividing the user into groups based on the basic information and the user's daily records; After the group is identified, make preliminary recommendations of relevant information at an appropriate time point; Based on the common interests of the same group, recommend products and commodity information commonly searched by the group to the users; Acquire stage data related to the user through other departments and platforms, and recommend suggestions and products associated with the stage data to the user; Grasp the information in advance, analyze the interest level of each group in the information, plan the aspects and products that the users are interested in in advance based on the information, and push the information.
[0006] In a possible implementation, dividing the groups according to the basic information and the daily records of the users includes: The basic information includes age, gender, region, education level, income, job, address, family members and hobbies.
[0007] In a possible implementation, dividing the groups according to the basic information and the daily records of the users includes: The user's daily records include shopping records, search records, bank records and platform records.
[0008] In a possible implementation, dividing the groups according to the basic information and the daily records of the users includes: By correlating and summarizing the basic information and daily records, the interests and hobbies of the user and the types and categories of the required products are determined.
[0009] In a possible implementation, the preliminary recommendation of relevant information at an appropriate time point after the group is determined includes: According to the division of the groups and the habits of the users, the time periods of the users' work, study, rest and entertainment are determined, and information is recommended in appropriate time periods.
[0010] In a possible implementation, the relying on the common interests of the same group includes: By analyzing and summarizing the group, the common interests and multiple intended products of the group are determined, and the multiple products of interest are identified based on the search records and information obtained by each member of the group, and finally recommendations are made to multiple users belonging to the same group.
[0011] In a possible implementation, the acquiring of the stage data related to the user through other departments and platforms includes: The user's current stage and plan are clarified through the feedback from departments and platforms. The stage data includes home purchase, car purchase, childbirth, medical care, education, travel and borrowing loans; based on this, relevant content is summarized and recommended products and suggestions are recommended.
[0012] In a possible implementation, the obtaining of information in advance includes: Search for news through the content fed back by the group and track and bind the relevant information; Implement and clarify the progress of the information and determine the recommended time points for products related to the information.
[0013] In a possible implementation, planning in advance the aspects and products that the user is interested in based on the information and pushing the information includes: Determine the degree of reaction of each group to the information and rank them, determine the interest points of each group to the information and different product information related to different interest points, and finally push different product information to different groups in different ways.
[0014] In a possible implementation, planning in advance the aspects and products that the user is interested in based on the information and pushing the information includes: Based on the analysis of each of the groups, different corresponding news and reports are pushed, and combined with the feedback level of the users, relevant product recommendations and related information are pushed.
[0015] The beneficial effect of the recommendation and push method based on information and neural network provided by the present invention is that: compared with the prior art, the recommendation and push method based on information and neural network provided by the present invention first obtains the basic information of the user with the consent of the user, divides the group according to the basic information and the daily records of the user, and makes preliminary recommendations of relevant information at an appropriate time after the group is determined.
[0016] In order to further provide users with more targeted product information, we can rely on the common interests of the same group, obtain user-related stage data through other departments and platforms, and grasp the information in advance, analyze the interest level of each group in the information, and plan the aspects and products that users are interested in in advance based on the information and push the information. This application can accurately provide users with suitable products and related information through the above method, with advanceness, can accurately lock in users' interests, improve the effect and accuracy of information recommendation, and can continuously update the recommended content according to the different stages of users. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0018] Figure 1 A flowchart of a recommendation push method based on information and neural network provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0019] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.
[0020] See also Figure 1 Now, the recommendation push method based on information and neural network provided by the present invention is described. The recommendation push method based on information and neural network includes: With the user's consent, the user's basic information is obtained, and groups are divided according to the basic information and the user's daily records.
[0021] After the group is determined, preliminary recommendations of relevant information will be made at an appropriate time.
[0022] Based on the common interests of the same group, we recommend to users products and commodity information that are commonly searched by the group.
[0023] Obtain user-related stage data through other departments and platforms, and recommend suggestions and products associated with the stage data to users.
[0024] Grasp the information in advance, analyze the interest level of each group in the information, plan the aspects and products that users are interested in in advance based on the information, and push the information.
[0025] The beneficial effect of the recommendation and push method based on information and neural network provided by the present invention is that: compared with the prior art, the recommendation and push method based on information and neural network provided by the present invention first obtains the basic information of the user with the consent of the user, divides the group according to the basic information and the daily records of the user, and makes preliminary recommendations of relevant information at an appropriate time after the group is determined.
[0026] In order to further provide users with more targeted product information, we can rely on the common interests of the same group, obtain user-related stage data through other departments and platforms, and grasp the information in advance, analyze the interest level of each group in the information, and plan the aspects and products that users are interested in in advance based on the information and push the information. This application can accurately provide users with suitable products and related information through the above method, with advanceness, can accurately lock in users' interests, improve the effect and accuracy of information recommendation, and can continuously update the recommended content according to the different stages of users.
[0027] In some embodiments of the recommendation push method based on information and neural network provided in the present application, the division of groups according to basic information and daily records of users includes: Basic information includes age, gender, region, education, income, job, address, family members and hobbies.
[0028] Due to its outstanding performance in improving the user experience of network applications and increasing network service traffic and revenue, the recommendation system has attracted widespread attention and interest from academia and industry. Specifically, collaborative filtering builds a user preference model and makes personalized recommendations by leveraging the user's past behavior, that is, the interaction between the user and the item. The user-item interaction is usually represented as a matrix, where a row corresponds to a user and a column corresponds to an item. The values in the matrix represent the user's preference for the item.
[0029] Traditional rating prediction algorithms only use historical ratings of users on items to model user preferences for personalized recommendations. However, due to data sparsity and cold start problems, the recommendation performance of matrix decomposition algorithms is limited. To alleviate these problems, social recommendation introduces social relationships between users when building user preference models. However, using only social relationships cannot properly characterize the similarities between users. With the rapid development of network technology, in addition to social relationships, more and more side information can be obtained through network services, such as user attributes, item attributes, text descriptions, etc.
[0030] In some embodiments of the recommendation push method based on information and neural network provided in the present application, the division of groups according to basic information and daily records of users includes: Users' daily records include shopping records, search records, banking records, and platform records.
[0031] With the rapid development of the Internet, the amount of data is growing exponentially. Due to the overload of information, it is difficult for users to select the content they are interested in from a large number of choices. In order to improve the user experience, recommendation systems have been applied to scenarios such as music recommendation, movie recommendation, and online shopping. However, traditional recommendation methods have the problem of data sparsity, such as item attributes, user profiles, and context, and therefore perform poorly in sparse situations where there are few interactions between users and items. In addition, traditional recommendation methods model each interaction as an independent data instance without considering the relationship between them. This makes them insufficient to extract attribute-based collaborative signals from the collective behavior of users. Therefore, to solve this problem, it is necessary to introduce auxiliary information as a supplement to the original data to alleviate the sparsity of the data and provide more high-order information to further enrich the connection between users and items.
[0032] Traditional information recommendation methods mostly recommend related products based on users' search records. However, users may not be interested in products after a word search, and repeated identical products cannot achieve good recommendation effects. Therefore, how to more intelligently classify various types of users to achieve good recommendation effects is a technical problem that needs to be solved urgently.
[0033] In some embodiments of the recommendation push method based on information and neural network provided in the present application, the division of groups according to basic information and daily records of users includes: By correlating and summarizing basic information and daily records, we can determine the user's interests and hobbies as well as the types and categories of products they need.
[0034] In order to solve the above problems, we first need to classify users of different types in different regions, so as to build a rough model of different users. After the model is created, we can then make targeted and direct product recommendations based on other users' search habits and product types. This process may not need to be based entirely on users' search habits and records, because the types of products and commodities that similar users pay attention to are similar. It is for the above reasons that the recommendation work can be completed more directly and simply.
[0035] To explain in more detail, each user is first classified according to the region, education level, identity, income, gender, personality, living habits and age based on the available information. After the classification is completed, preliminary specific recommendations can be made for the user. This method does not require the search history of a single user to complete the recommendation work.
[0036] In some embodiments of the recommendation and push method based on information and neural network provided in the present application, the preliminary recommendation of relevant information at an appropriate time point after the group is determined includes: According to the division of groups and user habits, the time periods of users for work, study, rest and entertainment are determined, and information is recommended at appropriate time periods.
[0037] It should be pointed out that the content that the same user is interested in will change greatly with different ages and living conditions. The degree of this change is relatively large. If the recommended content cannot attract the user's attention, it will eventually have the opposite effect.
[0038] Although existing technologies can analyze users' interests to a certain extent by building models and related algorithms, these interests are more related to the user's current search history and are not predictive. In other words, the user's preferred products may change greatly in a short period of time, and this change cannot be met by current algorithms.
[0039] Therefore, after classifying the users, it is also necessary to recommend products that may suit their needs in advance based on their different life stages and the acquisition of relevant information. The ultimate goal is to guide users to consume in advance or in a targeted manner, rather than for users to choose different products and search for the products they want themselves.
[0040] In some embodiments of the recommendation and push method based on information and neural network provided in the present application, relying on the common interests of the same group includes: By analyzing and summarizing the group, we can determine the group's common interests and multiple intended products. We can also combine the group's respective search records and information acquisition content to identify multiple products of interest, and ultimately make recommendations to multiple users in the same group.
[0041] After classifying different users, it should be pointed out that not all users will open the relevant shopping software and then search for products and commodities in the search box. It is precisely because of the above reasons that how to accurately and in advance provide customers with the information they may need to guide users to consume is a technical problem that needs to be focused on.
[0042] When users buy a house, they need information about decoration and other related products. When users buy a car, they need information about the car. In order to ensure accurate judgment of user intentions, it is necessary to plan the promotion of related products in advance based on the user's consumption. In order to achieve the above technical effects, it is necessary to search for consumption information related to the user that is not on this platform based on the user's consent. This information may exist in relevant departments or corresponding units, or other shopping platforms.
[0043] In some embodiments of the recommendation push method based on information and neural network provided in the present application, obtaining stage data related to the user through other departments and platforms includes: The user's current stage and plan are clarified through the feedback from departments and platforms. The stage data includes home purchase, car purchase, childbirth, medical care, education, travel and borrowing loans; based on this, relevant content is summarized and recommended products and suggestions are recommended.
[0044] To explain in more detail, the method in this application can create a relatively independent model for each user, and this model can accurately describe the consumption habits and focus of each user. More importantly, this model can produce corresponding changes based on the information obtained from various channels, that is, it can determine the possible points of interest of the user in the current time interval based on the information obtained.
[0045] On this basis, in order to further improve the predictability, since the life trajectories of most users are traceable, that is, most users have similar life situations at different ages, and it is precisely because of this convergence that users who belong to the same stage and have similar life situations can be divided into the same category, and then the purchase of goods and corresponding life patterns of this type of users can be recorded in real time. After the user category division is completed, then according to the user's acceptance of the recommended information and the acceptable frequency, the corresponding product information can be recommended to potential target users. That is, by characterizing the same group of people, according to the common characteristics of this group of people, and then according to the individual differences, the information is finally pushed.
[0046] In some embodiments of the recommendation push method based on information and neural network provided in this application, the advance grasp of information includes: Search for news and track related information through group feedback; Implement and clarify the progress of information and determine the recommended time points for products related to the information.
[0047] In this application, a large number of users have been classified, and the interests of this type of users have been summarized through classification. Traditional recommendation methods recommend the same products based on the user's current search habits. This method undoubtedly has a very large lag, and users usually will not pay attention to related products after searching once and understanding the related content. It is precisely because of the above reasons that the existing recommendation methods cannot complete the recommendation work well.
[0048] In order to solve this problem, the present application completes the characterization of this type of user after classifying each user. The important aspect of this is that it can not only accurately determine the current user's interest in the products, but also predict possible points of interest in the future.
[0049] In some embodiments of the recommendation push method based on information and neural network provided in the present application, planning in advance the aspects and products that users are interested in based on information and pushing information includes: Determine the degree of each group's reaction to the information and rank them, determine each group's interest points in the information and the different product information related to different interest points, and finally push different product information to different groups differently.
[0050] The existing eye capture technology uses corresponding instruments to accurately grasp the position of the eyeballs and ultimately determine the current focus of the person. The problem with this technology is that it requires the purchase of corresponding equipment and has fewer application scenarios. In order to solve this problem, this application detects the user's focus through real-time scanning of the front camera and corresponding sensors on the basis of user authorization. More importantly, it detects the user's expressions and movements in real time to determine the aspects that the user is interested in and how to obtain the intended information. Once the above situation is determined, it is possible to achieve differentiated processing of each user on the basis of classification, thereby facilitating the push of more accurate information.
[0051] In some embodiments of the recommendation push method based on information and neural network provided in the present application, planning in advance the aspects and products that users are interested in based on information and pushing information includes: Based on the analysis of each group, we push corresponding news and reports, and recommend related products and push relevant information based on the user's feedback.
[0052] That is, in this method, it is necessary to grasp the information that the corresponding user group may be interested in in real time, because this information will stimulate the interest of a specific user group and lead them to purchase the corresponding products. To this end, it is necessary to push information and recommend related products to specific groups of people at the right time based on the date and corresponding effects of the information in advance.
[0053] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A recommendation and push method based on information and neural network, characterized in that: include: Obtaining basic information of the user with the user's consent, and dividing the user into groups based on the basic information and the user's daily records; After the group is identified, make preliminary recommendations of relevant information at an appropriate time point; Based on the common interests of the same group, recommend products and commodity information commonly searched by the group to the users; Acquire stage data related to the user through other departments and platforms, and recommend suggestions and products associated with the stage data to the user; Grasp the information in advance, analyze the interest level of each group in the information, plan the aspects and products that the users are interested in in advance based on the information, and push the information.
2. The recommendation push method based on information and neural network as claimed in claim 1, characterized in that: The dividing of groups according to the basic information and the daily records of the users includes: The basic information includes age, gender, region, education level, income, job, address, family members and hobbies.
3. The recommendation push method based on information and neural network as claimed in claim 2, characterized in that: The dividing of groups according to the basic information and the daily records of the users includes: The user's daily records include shopping records, search records, bank records and platform records.
4. The recommendation push method based on information and neural network as claimed in claim 3, characterized in that: The dividing of groups according to the basic information and the daily records of the users includes: By correlating and summarizing the basic information and daily records, the interests and hobbies of the user and the types and categories of the required products are determined.
5. The recommendation push method based on information and neural network as claimed in claim 2, characterized in that: The preliminary recommendation of relevant information at an appropriate time point after the group is determined includes: According to the division of the groups and the habits of the users, the time periods of the users' work, study, rest and entertainment are determined, and information is recommended in appropriate time periods.
6. The recommendation push method based on information and neural network as claimed in claim 2, characterized in that: The said relying on the common interests of the same group includes: By analyzing and summarizing the group, the common interests and multiple intended products of the group are determined, and the multiple products of interest are identified based on the search records and information obtained by each member of the group, and finally recommendations are made to multiple users belonging to the same group.
7. The recommendation push method based on information and neural network as claimed in claim 2, characterized in that: The stage data related to the user obtained through other departments and platforms includes: The user's current stage and plan are clarified through the feedback from departments and platforms. The stage data includes home purchase, car purchase, childbirth, medical care, education, travel and loan borrowing; based on this, the relevant content is summarized and recommended products and suggestions are recommended.
8. The recommendation push method based on information and neural network as claimed in claim 2, characterized in that: The advance grasp of information includes: Search for news through the content fed back by the group and track and bind the relevant information; Implement and clarify the progress of the information and determine the recommended time points for products related to the information.
9. The recommendation push method based on information and neural network as claimed in claim 8, characterized in that: The planning of aspects and commodities of interest to the user in advance based on the information and pushing of information includes: Determine the degree of reaction of each group to the information and rank them, determine the interest points of each group to the information and different product information related to different interest points, and finally push different product information to different groups in different ways.
10. The recommendation push method based on information and neural network as claimed in claim 9, characterized in that: The planning of aspects and commodities of interest to the user in advance based on the information and pushing of information includes: Based on the analysis of each of the groups, different corresponding news and reports are pushed, and combined with the feedback level of the users, relevant product recommendations and related information are pushed.