Recommendation method and device based on user interest points

Abstract

The invention discloses a recommendation method and device based on user interest points, and relates to the technical field of interest point recommendation, and the method comprises the steps: S1, obtaining behavior data, time information and attributes of interactive items of a user on a platform, and carrying out the preprocessing of the obtained data; and S2, performing short-term preference analysis and long-term preference analysis based on the processed user data, modeling interest points of the user, generating a user portrait and the like. According to the method, short-term preference analysis and long-term preference analysis are respectively carried out by utilizing the user data, and a unified user portrait is formed by combining short-term and long-term preference characteristics, so that more accurate personalized recommendation is provided for the user, the user satisfaction and the scene matching degree are improved, and the user experience is improved through a long-term and short-term preference fusion strategy. Therefore, the recommendation system can utilize the existing resources more efficiently and obtain more comprehensive and abundant preference representation of the user, so that more efficient and accurate interest point recommendation is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of point of interest recommendation, and in particular to a method and device for recommending points of interest based on users. Background Art

[0002] With the rapid development of positioning technology, mobile smart devices and mobile Internet technology, location-based social networks have become more popular and have gradually become an important platform for people to check in and discover points of interest. Using a large amount of user historical check-in data to mine user habits and preferences and recommend new places that users may be interested in is called point of interest recommendation. Point of interest recommendation plays an important role in social networks.

[0003] At present, traditional POI recommendation methods often make static recommendations based on historical data and cannot adapt to changes in user interests and emerging trends in a timely manner. Moreover, user interests change dynamically, making it difficult to accurately judge user behavior needs and unable to obtain comprehensive user preference representations, resulting in poor recommendation results. Summary of the invention

[0004] The purpose of the present invention is to provide a recommendation method and device based on user interest points, so as to solve the problem proposed in the above background technology that the traditional interest point recommendation method is often based on historical data for static recommendation, and cannot adapt to the changes in user interests and emerging trends in time, and the user's interests are dynamically changing, it is difficult to accurately judge the user's behavioral needs, and it is impossible to obtain the user's comprehensive preference expression, resulting in poor recommendation effect.

[0005] To achieve the above object, the present invention provides the following technical solution: a recommendation method based on user interest points, comprising the following steps:

[0006] S1. Obtain the user's behavior data, time information and attributes of interactive items on the platform and pre-process the acquired data;

[0007] S2: Perform short-term and long-term preference analysis based on the processed user data, model the user's interests, and generate a user profile;

[0008] S3, using the long-term and short-term preference fusion recommendation algorithm to provide interest point recommendations for each user, and sort and filter the recommendation results according to the user's interests and behavior records;

[0009] S4. After the recommendation results are displayed, collect user feedback information and use the feedback data to regularly retrain or adjust the recommendation model;

[0010] S5. Display the recommendation results to the user through a visual interface.

[0011] Preferably, in step S1, the preprocessing includes removing invalid data, duplicate data and outliers, standardizing data in different formats, and extracting important features from user behavior and item information.

[0012] Preferably, in step S2, the process of the short-term preference analysis is as follows: firstly, the frequency of user behaviors within a short-term time range is counted, and the calculation formula is as follows:

[0013] f(u,i)=count(actions u,i );

[0014] Where: f(u,i) is the short-term preference frequency of user u for item i, actions u,i is the set of interaction records between user u and item i within the time window;

[0015] Assign different weights to actions in different periods, using weighted frequencies:

[0016]

[0017] Where T1 and T2 are the short-term and long-term time windows, and α(t) and β(t) are the weights of the corresponding time periods.

[0018] Preferably, in step S2, the process of long-term preference analysis is: statistically analyze the accumulated behavior data of users to identify long-term trends and patterns. The long-term preference of user u with item i in time period T can be represented by the total number of interactions:

[0019] F(u,i)=count(actions u,i,T );

[0020] Then calculate the proportion and frequency of each interest point in the long term to identify the user's main interest areas, and calculate the user's proportion in each interest category using the following formula:

[0021]

[0022] Where c is the interest category and C is the set of all categories.

[0023] Preferably, in step S3, providing interest point recommendations for each user using a long-term and short-term preference fusion recommendation algorithm includes the following steps:

[0024] S31. Combining short-term and long-term preference features:

[0025] profile(u)=λ·wf(u,i)+(1-λ)·F(u,i);

[0026] Where: λ is the weight coefficient of short-term preference and long-term preference;

[0027] S32, generating a user portrait based on all features to form a user's interest point vector;

[0028] S33. Generate a candidate item list based on the user portrait and item features.

[0029] Preferably, in step S3, the process of sorting and filtering the recommendation results is: scoring the candidate items, sorting them according to the item scores, selecting high-scoring items, filtering the results, and feeding back the final recommended item list to the customer.

[0030] Preferably, in step S4, the periodically retraining or adjusting the recommendation model using feedback data comprises the following steps:

[0031] S41, setting a feedback option on the recommendation result display page, and storing the user's feedback information in a structured form in the database;

[0032] S42, cleaning the collected feedback data, extracting features based on the feedback data, and weighting different feedback types;

[0033] S43, merging the obtained feedback data into the existing training data set, creating an incremental training set, and updating the user portrait and item feature matrix;

[0034] S44, using the updated data set to incrementally update the model training, adjusting the model parameters and hyperparameters to adapt to the new feedback information;

[0035] S45. Deploy the newly trained model to the recommendation system to replace the old model.

[0036] Preferably, in step S5, the process of presenting the recommendation results to the user through a visual interface is: using a grid or list view to present the recommended items according to the recommendation results, supporting user sliding and paging, and the user can quickly provide feedback during the interaction to improve the next recommendation.

[0037] A recommendation device based on user interest points, the device comprising:

[0038] Data collection module, used to collect user behavior data, points of interest and preference information;

[0039] Point of interest analysis module, used to analyze users' points of interest, identify and model users' short-term and long-term preferences;

[0040] The long-term and short-term fusion recommendation module is used to integrate the user's short-term preferences with the long-term preferences to generate comprehensive and rich preference recommendation results;

[0041] Feedback module, used for users to provide feedback on recommendation results and use feedback data to optimize and adjust the recommendation model;

[0042] The sending module is used to send the recommended points of interest to the user terminal.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] 1. In the present invention, short-term preference analysis and long-term preference analysis are performed separately when using user data, and the characteristics of short-term and long-term preferences are combined to form a unified user portrait, so as to provide users with more accurate personalized recommendations, improve user satisfaction and scene matching, and through the fusion strategy of long-term and short-term preferences, the recommendation system can more efficiently utilize existing resources and obtain a more comprehensive and rich preference representation of users, thereby achieving more efficient and accurate recommendation of points of interest.

[0045] 2. In the present invention, by continuously collecting user feedback and incremental learning, the obtained feedback data is merged into the existing training data set, an incremental training set is created, and the user portrait and item feature matrix are updated, so as to adapt to the user's changing consumption needs and interests in real time, ensure the relevance of the recommendation, and set a feedback mechanism and visual display to enhance the user's sense of participation, encourage users to provide feedback, and promote the interaction between users and the platform.

[0046] 3. In the present invention, the recommendation model is incrementally updated and optimized based on updated data, the newly trained model is deployed into the recommendation system to replace the old model, and the user portrait is regularly updated to make it more comprehensive and dynamic, so as to better reflect the user's real preferences and behaviors. The recommendation model is continuously optimized by effectively utilizing user feedback data to improve the accuracy of recommendations and the overall user satisfaction, and the recommendation system can achieve self-enhancement to ensure that it can always adapt to user behavior changes and preference updates. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 The present invention is a flowchart of a recommendation method based on user interest points. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the implementation regulations described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] See also Figure 1The present invention provides a technical solution: a recommendation method based on user interest points, comprising the following steps:

[0050] Step 1: Obtain the user's behavior data, time information, and attributes of interactive items on the platform and pre-process the acquired data; the pre-processing includes removing invalid data, duplicate data, and outliers, standardizing data in different formats, extracting important features from user behavior and item information, and providing basic data for subsequent analysis;

[0051] Step 2: Perform short-term and long-term preference analysis based on the processed user data, model the user's interests, and generate a user profile;

[0052] The process of short-term preference analysis is as follows: First, count the frequency of user behaviors within a short period of time. The calculation formula is as follows:

[0053] f(u,i)=count(actions u,i );

[0054] Where: f(u,i) is the short-term preference frequency of user u for item i, actions u,i is the set of interaction records between user u and item i within the time window;

[0055] Assign different weights to actions in different periods, using weighted frequencies:

[0056]

[0057] Where T1 and T2 are short-term and long-term time windows, α(t) and β(t) are the weights of the corresponding time periods;

[0058] The process of long-term preference analysis is as follows: statistical analysis of the user's accumulated behavior data is performed to identify long-term trends and patterns. The long-term preference of user u with item i in time period T can be represented by the total number of interactions:

[0059] F(u,i)=count(actions u,i,T );

[0060] Then calculate the proportion and frequency of each interest point in the long term to identify the user's main interest areas, and calculate the user's proportion in each interest category using the following formula:

[0061]

[0062] In the formula, c is the interest category, and C is the set of all categories;

[0063] Step 3: Use the long-term and short-term preference fusion recommendation algorithm to provide interest point recommendations for each user, and sort and filter the recommendation results according to the user's interests and behavior records. By combining short-term and long-term preferences, more accurate personalized recommendations are provided to users, improving user satisfaction and scene matching;

[0064] Among them, using the long-term and short-term preference fusion recommendation algorithm to provide interest point recommendations for each user includes the following steps:

[0065] 31) Combining short-term and long-term preference features:

[0066] profile(u)=λ·wf(u,i)+(1-λ)·F(u,i);

[0067] Where: λ is the weight coefficient of short-term preference and long-term preference;

[0068] 32) Generate a user portrait based on all features to form a user's interest point vector;

[0069] 33) Generate a list of candidate items based on user portraits and item features;

[0070] Specifically, the process of sorting and filtering the recommendation results is: scoring the candidate items, sorting them according to the item scores, selecting high-scoring items, filtering the results, and feeding back the final recommended item list to the customer;

[0071] Step 4: After the recommendation results are displayed, user feedback information is collected and the recommendation model is regularly retrained or adjusted using the feedback data. By continuously collecting user feedback and incremental learning, the system can adapt to the changing consumption needs and interests of users in real time and ensure the relevance of recommendations. The user feedback data can be effectively used to continuously optimize the recommendation model to improve the accuracy of recommendations and the overall satisfaction of users, realize the self-enhancement of the recommendation system, and ensure that it can always adapt to changes in user behavior and updates in preferences;

[0072] Among them, using feedback data to regularly retrain or adjust the recommendation model includes the following steps:

[0073] 41) Setting feedback options on the recommendation result display page and storing user feedback information in a structured form in the database;

[0074] 42) Clean the collected feedback data, extract features based on the feedback data, and weight different feedback types;

[0075] 43) Merge the obtained feedback data into the existing training data set, create an incremental training set, and update the user profile and item feature matrix;

[0076] 44) Use the updated data set to incrementally update the model and adjust the model parameters and hyperparameters to adapt to the new feedback information;

[0077] 45) Deploy the newly trained model to the recommendation system to replace the old model, and regularly update the user profile to make it more comprehensive and dynamic, so as to better reflect the user's real preferences and behaviors;

[0078] Step 5: Display the recommendation results to the user through a visual interface. The specific process is as follows: Use a grid or list view to display recommended items based on the recommendation results, support user sliding and paging, and users can quickly provide feedback during the interaction to improve the next recommendation. By setting up a feedback mechanism and visual display, the user's sense of participation is enhanced, and users are encouraged to provide feedback, thereby promoting interaction between users and the platform.

[0079] A recommendation device based on user interest points, the device comprising:

[0080] Data collection module, used to collect user behavior data, points of interest and preference information;

[0081] Point of interest analysis module, used to analyze users' points of interest, identify and model users' short-term and long-term preferences;

[0082] The long-term and short-term fusion recommendation module is used to integrate the user's short-term preferences with the long-term preferences to generate comprehensive and rich preference recommendation results;

[0083] Feedback module, used for users to provide feedback on recommendation results and use feedback data to optimize and adjust the recommendation model;

[0084] The sending module is used to send the recommended points of interest to the user end.

[0085] In the present invention, by obtaining the user's behavioral data, time information and attributes of interactive items on the platform, and using the user data to perform short-term preference analysis and long-term preference analysis respectively, a unified user portrait is formed by combining the characteristics of short-term and long-term preferences, providing users with more accurate personalized recommendations, improving user satisfaction and scene matching, and by continuously collecting user feedback and incremental learning, the system can adapt to the user's changing consumption needs and interests in real time to ensure the relevance of recommendations. By setting a feedback mechanism and visual display, the user's sense of participation is enhanced, and users are encouraged to provide feedback, thereby promoting interaction between users and the platform. Through the fusion strategy of long-term and short-term preferences, the recommendation system can more efficiently utilize existing resources and obtain a more comprehensive and rich preference representation of users, thereby achieving more efficient and accurate recommendation of points of interest.

[0086] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. 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 method based on user interest points, characterized in that: The following steps are involved: S1. Obtain the user's behavior data, time information and attributes of interactive items on the platform and pre-process the acquired data; S2: Perform short-term and long-term preference analysis based on the processed user data, model the user's interests, and generate a user profile; S3, using the long-term and short-term preference fusion recommendation algorithm to provide interest point recommendations for each user, and sort and filter the recommendation results according to the user's interests and behavior records; S4. After the recommendation results are displayed, collect user feedback information and use the feedback data to regularly retrain or adjust the recommendation model; S5. Display the recommendation results to the user through a visual interface.

2. The method for recommending based on user interest points according to claim 1, characterized in that: In step S1, the preprocessing includes removing invalid data, duplicate data and outliers, standardizing data in different formats, and extracting important features from user behavior and item information.

3. The method for recommending based on user interest points according to claim 1, characterized in that: In step S2, the process of the short-term preference analysis is as follows: firstly, the frequency of user behaviors within a short-term time range is counted, and the calculation formula is as follows: f(u,i)=count(actionsu,i); Where: f(u,i) is the short-term preference frequency of user u for item i, actions u,i is the set of interaction records between user u and item i within the time window; Assign different weights to actions in different periods, using weighted frequencies: Where T1 and T2 are the short-term and long-term time windows, and α(t) and β(t) are the weights of the corresponding time periods.

4. The method for recommending based on user interest points according to claim 1, characterized in that: In step S2, the process of long-term preference analysis is: statistical analysis is performed on the user's accumulated behavior data to identify long-term trends and patterns. The long-term preference of user u with item i in time period T can be represented by the total number of interactions: F(u,i)=count(actionsu,i, T ); Then calculate the proportion and frequency of each interest point in the long term to identify the user's main interest areas, and calculate the user's proportion in each interest category using the following formula: Where c is the interest category and C is the set of all categories.

5. The method for recommending based on user interest points according to claim 1, characterized in that: In step S3, the method of providing interest point recommendations for each user using the long-term and short-term preference fusion recommendation algorithm includes the following steps: S31. Combining short-term and long-term preference features: profile(u)=λ·wf(u,i)+(1-λ)·F(u,i); Where: λ is the weight coefficient of short-term preference and long-term preference; S32, generating a user portrait based on all features to form a user's interest point vector; S33. Generate a candidate item list based on the user portrait and item features.

6. The method for recommending based on user interest points according to claim 1, characterized in that: In step S3, the process of sorting and filtering the recommendation results is: scoring the candidate items, sorting them according to the item scores, selecting high-scoring items, filtering the results, and feeding back the final recommended item list to the customer.

7. The method for recommending based on user interest points according to claim 1, characterized in that: In step S4, the periodic retraining or adjustment of the recommendation model using feedback data includes the following steps: S41, setting a feedback option on the recommendation result display page, and storing the user's feedback information in a structured form in the database; S42, cleaning the collected feedback data, extracting features based on the feedback data, and weighting different feedback types; S43, merging the obtained feedback data into the existing training data set, creating an incremental training set, and updating the user portrait and item feature matrix; S44, using the updated data set to incrementally update the model training, adjusting the model parameters and hyperparameters to adapt to the new feedback information; S45. Deploy the newly trained model to the recommendation system to replace the old model.

8. The method for recommending based on user interest points according to claim 1, characterized in that: In step S5, the process of presenting the recommendation results to the user through a visual interface is as follows: based on the recommendation results, the recommended items are presented in a grid or list view, and user sliding and paging are supported. During the interaction, the user can provide quick feedback to improve the next recommendation.

9. A recommendation device based on user interest points, characterized in that: The device comprises: Data collection module, used to collect user behavior data, points of interest and preference information; Point of interest analysis module, used to analyze users' points of interest, identify and model users' short-term and long-term preferences; The long-term and short-term fusion recommendation module is used to integrate the user's short-term preferences with the long-term preferences to generate comprehensive and rich preference recommendation results; Feedback module, used for users to provide feedback on recommendation results and use feedback data to optimize and adjust the recommendation model; The sending module is used to send the recommended points of interest to the user terminal.

Images