Big data analysis processing method and system for personalized recommendation

Abstract

The invention relates to the technical field of data processing, and particularly discloses a big data analysis processing method for personalized recommendation. The user information of the registered user and the current hot product information of the platform can be acquired, the recommendation priority grading data can be comprehensively acquired, the periodic recommendation database can be established through combined analysis, the to-be-recommended information can be acquired according to the periodic recommendation database, and the to-be-recommended information can be acquired according to the comparative analysis of the to-be-recommended information and the actual purchase data. According to the method, recommendation time and content can be planned according to the characteristics of the user, the accepting rhythm of the user is not intensively and frequently pushed and respected in a short time, the user is prevented from being excessively disturbed, the user experience is improved, and the user experience is improved. In addition, a period recommendation database can be compared and updated in real time according to actual purchase data, user demand changes can be dynamically tracked, recommendation strategies are adjusted according to consumption behaviors, and the conflict psychology caused by improper pushing is reduced.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a big data analysis and processing method for personalized recommendation and a system thereof. Background Art

[0002] In today's digital age, personalized recommendations have been widely used in many fields such as e-commerce, online entertainment, content information, etc. Its purpose is to provide users with product or service recommendations that meet their interests and needs through analysis of user data, so as to enhance user experience, increase user engagement and promote business growth.

[0003] However, most existing personalized recommendation systems lack an effective real-time feedback mechanism and are unable to capture users’ actual reactions to recommended content in a timely manner. After users receive uninteresting or even wrong recommendation information, the system cannot immediately perceive and adjust subsequent recommendation strategies. For example, users have explicitly rejected recommendations for certain types of electronic products many times, but the system continues to push related products because it does not receive timely feedback, which not only wastes users’ time but also causes users’ disgust.

[0004] In pursuit of high exposure, some recommendation systems frequently push a large amount of recommended information to users without considering the user's acceptance and actual needs. Existing technologies usually obtain user recommendation information through big data analysis, and send promotional emails or push notifications to users multiple times in a short period of time, which seriously interferes with the user's normal use of products or services, making them feel overly disturbed, and thus resisting the recommended information. Summary of the invention

[0005] The object of the present invention is to provide a big data analysis and processing method for personalized recommendation, a big data analysis and processing method for personalized recommendation, comprising: Acquire user information of a registered user, wherein the user information includes preference feature data and taboo feature data; Get the current hot product information on this platform; Comparing and analyzing the preference feature data with the current popular product information, obtaining recommendation priority classification data; Performing combined analysis based on the recommendation priority classification data and the taboo feature data, and establishing a periodic recommendation database; Acquiring proposed recommendation information according to the periodic recommendation database; Get the actual purchase data of registered users; Based on the comparison and analysis between the proposed recommendation information and the actual purchase data, the periodic recommendation database is updated in real time; Send optimized recommendation information to corresponding registered users based on the updated periodic recommendation database.

[0006] Furthermore, the step of obtaining the user information of the registered user includes: By connecting the data interface with the data collection tool and following the data acquisition rules and privacy policy of this platform, we can obtain the user's behavior data on social media; Convert and standardize the acquired behavior data to make it compatible with the data structure of this platform; Correct the processed behavioral data, remove erroneous data, and obtain preference feature data; By analyzing the return and exchange reason records of users after purchase, we can extract information related to the product being unsuitable or not meeting expectations and obtain taboo feature data.

[0007] Furthermore, the step of obtaining the recommendation priority classification data includes: Create a two-dimensional matrix, where the rows represent the user's preference data features and the columns represent the popular product information; According to the matching degree between the user preference feature data and the popular product information attributes, the corresponding score is filled in the matrix. If the user's preferred product category completely matches the popular product category, the score is 10, if it partially matches, the score is 5; if it does not match, the score is 0; For each popular product, calculate its score in terms of preference matching, which is calculated as: ; in, is the preference matching score of the ith popular product, is the weight of the jth preference feature, is the score of user preference feature j and popular product i in the matching matrix, and n is the number of preference features; Taking into account the priority indicators of popular products, we set weights for each indicator, where the weight of sales volume is 0.4, the weight of sales revenue is 0.3, the weight of browsing popularity is 0.2, and the weight of collection volume is 0.1; The priority score calculation formula is: ; in, is the popularity score of the ith popular product, is the standardized value of sales volume of the ith product, is the normalized value of sales, is the normalized value of browsing popularity, is the standardized value of collection quantity; According to the priority level of different products Sort in descending order to obtain recommended priority grading data.

[0008] Furthermore, the step of acquiring the proposed recommendation information according to the periodic recommendation database includes: Collect and integrate user preference feature data, taboo feature data, recommendation priority classification data, and various product-related information to build a complete data set; Based on the taboo feature data, check the products in the recommendation priority classification data one by one, eliminate products containing taboo features, and ensure that the recommended products meet the basic requirements of users; Analyze user behavior and product usage cycle patterns, and determine the recommended time node or time range in the cycle recommendation database for the screened products based on the current time, so as to achieve accurate time planning and establish a cycle recommendation database; Monitor the user's contextual information and recent behavior trends in real time, match them with the product information in the periodic recommendation database, and select products that match the user's current situation and interests; Based on the matching results, personalized customization is performed according to the user's device and usage scenario, and the selected product information is converted into recommended information suitable for display.

[0009] Furthermore, the step of obtaining the actual purchase data of the registered user includes: Obtain purchase records of registered users extracted from the transaction database, including details of purchased products or services, purchase time, purchase amount, and payment method; Filter out valid and complete purchase data, and exclude test data, error records, and incomplete transaction information; Clean the extracted purchase data, deal with noise and outliers in the data, correct obviously erroneous price data, and unify the date; Integrate purchase data with other relevant information of the user, associate data from different sources through the user's unique identifier, and build a complete user purchase behavior data set; Store the processed actual purchase data in the data analysis storage area or database table, establish a data update mechanism, and obtain the latest purchase data regularly.

[0010] Furthermore, the step of sending the optimization recommendation information to the corresponding registered user according to the updated periodic recommendation database includes: Continuously obtain new actual purchase data of registered users, including purchase product information, purchase time, purchase quantity, etc., and organize it into a structured format; Extract the historical recommendation records of relevant products and the corresponding recommendation time, recommendation strategy and other information from the periodic recommendation database, and conduct a comprehensive comparative analysis; Compare the proposed recommendation information with the actual purchase data in detail to determine whether the product actually purchased by the user is within the proposed recommendation range, and the difference between the purchased product and the recommended product; Based on the user's purchasing behavior, feedback data and calculated deviation value, clustering algorithms or classification models are used to divide users into different groups; such as stable purchasing behavior type, variable purchasing behavior type, recommendation-sensitive type, recommendation-insensitive type, etc.

[0011] Analyze the behavior patterns and preference change trends of different user groups to identify common patterns that lead to recommendation bias; for example, a certain type of user prefers low-priced products during a promotion, and the recommendation does not take price factors into consideration, resulting in deviation.

[0012] Based on the results of user group classification and behavior pattern recognition, formulate targeted periodic recommendation database update strategies for different groups; for users with stable purchasing behavior, the recommendation cycle can be appropriately extended and the recommended content can be fine-tuned; for users who are sensitive to recommendations, recommended products are updated in a timely manner according to their recent changes in interests.

[0013] Apply the formulated update strategy to the periodic recommendation database, update product recommendation information, recommendation time plan and other contents in real time, and send optimized recommendation information of the updated periodic recommendation database to the corresponding registered users.

[0014] Furthermore, the present invention also discloses a big data analysis and processing system for personalized recommendation, comprising: A first acquisition module, used to acquire user information of a registered user, wherein the user information includes preference feature data and taboo feature data; The second acquisition module is used to obtain the current popular product information of this platform; A first analysis module is used to compare and analyze the preference feature data with the current popular product information to obtain recommendation priority classification data; Establishing a module for combining and analyzing the recommendation priority classification data with the taboo feature data and establishing a periodic recommendation database; A third acquisition module is used to acquire the proposed recommendation information according to the periodic recommendation database; The fourth acquisition module is used to obtain the actual purchase data of the registered user; The second analysis module is used to update the periodic recommendation database in real time based on the comparison and analysis between the proposed recommendation information and the actual purchase data; The recommendation module is used to send optimized recommendation information to corresponding registered users according to the updated periodic recommendation database.

[0015] Furthermore, the first acquisition module includes: A first acquisition unit is used to acquire the user's behavior data on social media by connecting the data interface with the data collection tool and in accordance with the data acquisition rules and privacy policy of this platform; A processing unit, used to convert and standardize the acquired behavior data to make it compatible with the data structure of this platform; A correction unit, used to correct the processed behavior data, remove erroneous data therein, and obtain preference feature data; The second acquisition unit is used to analyze the return and exchange reason records of users after purchase, extract information related to the product being unsuitable or not meeting expectations, and acquire taboo feature data.

[0016] The present application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0017] The present application also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above method when executed by a processor.

[0018] The beneficial effects of this application are: Firstly, the present invention can establish a periodic recommendation database and plan the recommendation time and content according to the characteristics of the user. Instead of pushing in a concentrated and frequent manner within a short period of time, the invention respects the user's acceptance rhythm and avoids excessively disturbing the user. This is in sharp contrast to the prior art of multiple pushes within a short period of time. The present invention can also compare and update the periodic recommendation database in real time based on actual purchase data, dynamically track changes in user needs, adjust recommendation strategies according to their latest consumption behaviors, ensure that recommendations always match the user's current acceptance ability and actual situation, and reduce resistance caused by inappropriate pushes.

[0019] Secondly, the present invention can send optimized recommendation information based on the updated periodic recommendation database. Instead of simply sending the same recommendation repeatedly, more appropriate content is pushed after continuous optimization. This not only ensures that users receive valuable recommendations, but also avoids meaningless information bombardment, thereby improving users' favorability towards recommended information and overcoming the problem in the prior art of users resisting recommendations due to excessive disturbances. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A schematic diagram of a method flow chart proposed in an embodiment of the present application; The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0021] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0022] like Figure 1 As shown, the present application provides a big data analysis and processing method for personalized recommendation, including: S1. Obtain user information of a registered user, wherein the user information includes preference feature data and taboo feature data; S2. Obtain the current popular product information on this platform; S3, performing comparative analysis based on the preference feature data and the current popular product information to obtain recommendation priority classification data; S4, combining and analyzing the recommendation priority classification data and the taboo feature data, and establishing a periodic recommendation database; S5. Acquire proposed recommendation information according to the periodic recommendation database; S6. Obtaining actual purchase data of registered users; S7. Based on the comparison and analysis between the proposed recommendation information and the actual purchase data, the periodic recommendation database is updated in real time; S8. Send optimization recommendation information to corresponding registered users according to the updated periodic recommendation database.

[0023] As described in the above steps S1-S8, the present invention can establish a periodic recommendation database, plan the recommendation time and content according to the characteristics of the user, rather than push in a concentrated and frequent manner in a short period of time, respect the user's acceptance rhythm, and avoid excessively disturbing the user. This is in sharp contrast to the prior art of multiple pushes in a short period of time. The periodic recommendation database can also be compared and updated in real time based on actual purchase data, and can dynamically track changes in user needs, adjust recommendation strategies according to their latest consumption behaviors, ensure that recommendations always meet the user's current acceptance ability and actual situation, and reduce resistance caused by inappropriate pushes.

[0024] The present invention can send optimized recommendation information based on the updated periodic recommendation database. Instead of simply sending the same recommendation repeatedly, more appropriate content is pushed after continuous optimization. This not only ensures that users receive valuable recommendations, but also avoids meaningless information bombardment, thereby improving users' favorability towards recommended information and overcoming the problem in the prior art of users resisting recommendations due to excessive disturbances.

[0025] In general, the present invention builds a recommendation mechanism that better meets users' personalized needs, respects users' acceptance capabilities, and continuously optimizes through in-depth analysis and utilization of users' multi-dimensional data. It effectively solves the problem of user resistance caused by excessive push in existing recommendation systems, and improves user experience and the effectiveness of recommendations.

[0026] Specifically, the steps of obtaining the user information of the registered user include: S9. Obtain user behavior data on social media by connecting the data interface to the data collection tool and in accordance with the data acquisition rules and privacy policy of this platform; S10, converting and standardizing the acquired behavior data to make it compatible with the data structure of this platform; S11, correcting the processed behavior data, removing erroneous data therein, and obtaining preference feature data; S12. Analyze the return and exchange reason records after purchase by users, extract information related to the product being unsuitable or not meeting expectations, and obtain taboo feature data.

[0027] As described in the above steps S9-S12, by connecting the data interface with the data collection tool, the user's behavior data on social media is obtained, which broadens the channels for data collection. Social media contains rich user behavior information, which can more comprehensively reflect the user's interests, preferences, and daily concerns, etc., and help to build a more three-dimensional and accurate user portrait. Compared with the traditional method that only relies on the platform's internal data, it can greatly improve the depth and breadth of understanding of users, thereby making subsequent recommendations more in line with the user's real needs.

[0028] The acquired behavioral data is formatted and standardized to make it compatible with the data structure of this platform, ensuring that data from different sources can be effectively integrated and analyzed. This avoids data processing problems caused by inconsistent data formats, improves the efficiency of data utilization, makes the subsequent analysis process smoother, and ensures that the entire personalized recommendation process can be carried out smoothly based on high-quality, collaborative data. The processed behavioral data is corrected to remove erroneous data, which helps to improve the accuracy and reliability of the data. The existence of erroneous data may interfere with the judgment of user preferences and other characteristics. The data can be purified through the correction link. The preference feature data obtained on this basis will be more accurate, providing strong support for subsequent accurate recommendations and reducing recommendation deviations caused by erroneous data.

[0029] Extracting taboo feature data from the return and exchange reason records after purchase is an effective way to mine key information from actual negative feedback from users. This method can accurately capture the product-related features that users clearly dislike and are not suitable for, making the constructed taboo feature data highly targeted, so that in the recommendation process, these products or services that users dislike can be effectively avoided, greatly improving the user's acceptance of the recommended content, avoiding user dissatisfaction caused by recommending content that users taboo, and optimizing the user experience.

[0030] Specifically, the step of obtaining the recommendation priority classification data includes: S13. Create a two-dimensional matrix, where the rows represent the user's preference data features and the columns represent the popular product information; S14. Fill in the corresponding score in the matrix according to the matching degree between the user preference feature data and the popular product information attributes, wherein if the user's preferred product category completely matches the popular product category, the score is 10, if it partially matches, the score is 5; if it does not match, the score is 0; S15. For each popular product, calculate its score in terms of preference matching. The calculation formula is: ; in, is the preference matching score of the ith popular product, is the weight of the jth preference feature, is the score of user preference feature j and popular product i in the matching matrix, and n is the number of preference features; S16. Comprehensively consider the priority indicators of popular products and set weights for each indicator, among which the weight of sales volume is 0.4, the weight of sales revenue is 0.3, the weight of browsing popularity is 0.2, and the weight of collection volume is 0.1; The priority score calculation formula is: ; in, is the popularity score of the ith popular product, is the standardized value of sales volume of the ith product, is the normalized value of sales, is the normalized value of browsing popularity, is the standardized value of collection quantity; S17. According to the priority level of different products Sort in descending order to obtain recommended priority grading data.

[0031] As described in the above steps S10-S17, the matching relationship between the abstract user preference feature data and the popular product information is quantified by creating a two-dimensional matrix and setting matching scores. This intuitive quantification method can clearly show the degree of fit between each user preference and each popular product, changing the previous practice of measuring matching conditions based solely on subjective judgment or fuzzy standards, making subsequent analysis more scientific, objective and operational.

[0032] The present invention also comprehensively considers the weights of different preference features and the specific scores of the matches, making the measurement of the degree of user preference matching more accurate and detailed. It can comprehensively evaluate the compatibility of products with user preferences from multiple dimensions, avoiding the one-sidedness of single factor judgment, and helping to screen out products that truly meet user interests and enter the recommendation list.

[0033] We comprehensively consider multiple priority indicators of popular products (sales volume, sales, browsing popularity, and collection volume), and set reasonable weights for them, comprehensively covering the product's market performance and user attention. This means that we do not judge the recommendation priority of a product based on just one indicator, but rather conduct a comprehensive analysis from multiple angles, so that the product priority situation can better reflect its true popularity and attention level, making the basis for recommendation more sufficient and reliable.

[0034] By arranging products in descending order according to their priority to obtain recommendation priority classification data, many products can be arranged in order according to the comprehensive situation of product matching with user preferences and popularity. In this way, when making recommendations, products with high scores and strong priorities can be pushed to users first, ensuring that the recommended products are more targeted and valuable, avoiding disordered recommendations that cause users to be faced with a large amount of chaotic recommendation information and have no choice, effectively improving the accuracy and effectiveness of recommendations, enhancing user experience, and optimizing the configuration of recommendation resources.

[0035] Specifically, the step of acquiring the proposed recommendation information according to the periodic recommendation database includes: S18. Collect and integrate user preference feature data, taboo feature data, recommendation priority classification data, and various product related information to build a complete data set; S19. Based on the taboo feature data, check the products in the recommendation priority classification data one by one, eliminate the products containing taboo features, and ensure that the recommended products meet the basic requirements of users; S20, analyzing user behavior and product usage cycle rules, and combining the current time to determine the recommended time node or time range in the cycle recommendation database for the screened products, to achieve accurate time planning, and to establish a cycle recommendation database; S21, monitor the user's context information and recent behavior trends in real time, match them with the product information in the periodic recommendation database, and select products that match the user's current situation and interests; S22. Based on the matching results, personalized customization is performed according to the user's device and usage scenario, and the selected product information is converted into proposed recommended information suitable for display.

[0036] As described in the above steps S18-S22, a complete data set is constructed by collecting and integrating multi-dimensional data, covering user preference features, taboo features, recommendation priority levels, and various product-related information. This comprehensive data integration method can provide a rich and detailed data foundation for the subsequent recommendation process, avoiding inaccurate recommendations due to missing or one-sided data, and helping to more comprehensively grasp user needs and product conditions as a whole, creating good conditions for accurate recommendations.

[0037] By deeply analyzing user behavior and product usage cycle patterns, and combining the current time to determine the recommendation time node or range, this precise time planning has obvious advantages. It changes the indiscriminate and random push mode of traditional recommendations, and arranges the recommendation time according to the user's actual consumption rhythm and the characteristics of the product itself. It will not make users feel that the recommendation is redundant too early, nor will it miss the user's demand period too late. It ensures that the recommendation can reach the user at the right time, in line with the user's usage habits, improves the timeliness and effectiveness of the recommendation, and avoids disturbing users due to inappropriate push.

[0038] The present invention fully considers the user's current dynamic situation by monitoring the user's context information and recent behavior trends in real time and matching them with the product information in the periodic recommendation database. Since the user's interests and needs may change at any time, this real-time matching can capture these changes in time and flexibly adjust the recommended content so that the recommendation always keeps up with the user's latest focus. Compared with static and fixed recommendations, it is more timely and targeted, allowing users to feel that the recommendation is in line with their current situation, thereby increasing the user's attention and favorability to the recommendation.

[0039] Specifically, the step of obtaining the actual purchase data of the registered user includes: S23, obtaining the purchase records of the registered user extracted from the transaction database, including the purchased product or service details, purchase time, purchase amount, and payment method; S24. Filter out valid and complete purchase data, and exclude test data, error records, and incomplete transaction information; S25. Clean the extracted purchase data, process the noise and outliers in the data, correct the obviously erroneous price data, and unify the date; S26, integrating the purchase data with other relevant information of the user, associating data from different sources through the user's unique identifier, and constructing a complete user purchase behavior data set; S27. Store the processed actual purchase data in the data analysis storage area or database table, establish a data update mechanism, and regularly obtain the latest purchase data.

[0040] As described in the above steps S23-S27, the present invention can extract the multi-dimensional purchase records of registered users from the transaction database, covering key contents such as product or service details, purchase time, purchase amount, payment method, etc. Such targeted extraction can obtain the core data reflecting the actual consumption behavior of users, and provide the most basic and direct basis for the subsequent in-depth analysis of user purchase preferences, consumption capacity, consumption cycle, etc., which helps to accurately grasp the real purchase characteristics of users, which is different from the general and unfocused data acquisition method, so that the entire personalized recommendation has a solid data starting point.

[0041] By screening valid and complete purchase data, and actively excluding test data, erroneous records, and incomplete transaction information, data can be effectively purified. In a big data environment, raw data often has various quality issues, and this step can eliminate "impurity" data that interferes with the accuracy of analysis, ensuring that subsequent processing and analysis are based on high-quality data, improving data reliability, and laying a solid foundation for accurate personalized recommendations, avoiding biased recommendations due to low-quality data.

[0042] Integrate the purchase data with other relevant information of the user, and associate data from different sources through the user's unique identifier to build a complete user purchase behavior data set. This approach breaks the data silos, fully explores the correlation between data, and can analyze user behavior from a more comprehensive perspective. For example, combined with the user's browsing history, search history and other information, we can have a deeper understanding of the motivations and influencing factors behind the user's purchase decision, so that personalized recommendations are no longer limited to a single purchase data, but are based on the user's overall behavior portrait, which greatly improves the comprehensiveness and accuracy of the recommendation, and is more in line with the complex and changing actual needs of users. The actual purchase data after processing is stored, and a data update mechanism is established to regularly obtain the latest purchase data to ensure the timeliness of the data. The user's purchase behavior is dynamically changing. Regularly updating data can enable the recommendation system to keep abreast of the user's latest consumption trends, so that the recommended content can be adjusted accordingly according to the user's current purchase situation, avoiding the situation where the recommendation does not meet the user's current needs due to outdated data, ensuring that the recommendation can always keep up with the pace of changes in user behavior, enhancing the adaptability and effectiveness of the recommendation, and improving user satisfaction with the recommendation service.

[0043] Specifically, the step of sending the optimization recommendation information to the corresponding registered user according to the updated periodic recommendation database includes: S28. Continuously obtain new actual purchase data of registered users, including purchase product information, purchase time, purchase quantity, etc., and organize them into a structured format; S29, extracting historical recommendation records of relevant products and corresponding recommendation time, recommendation strategy and other information from the periodic recommendation database, and conducting a comprehensive comparative analysis; S30, comparing the proposed recommendation information with the actual purchase data in detail to determine whether the product actually purchased by the user is within the proposed recommendation range, and the difference between the purchased product and the recommended product; S31. Based on the user's purchasing behavior, feedback data and calculated deviation value, a clustering algorithm or classification model is used to divide the users into different groups; such as stable purchasing behavior type, variable purchasing behavior type, recommendation-sensitive type, recommendation-insensitive type, etc.

[0044] S32. Analyze the behavior patterns and preference change trends of different user groups, and identify common patterns that lead to recommendation bias; for example, a certain type of user prefers low-priced products during a promotion period, and the recommendation does not take price factors into consideration, resulting in deviation.

[0045] S33. Based on the results of user group classification and behavior pattern recognition, formulate targeted periodic recommendation database update strategies for different groups; for users with stable purchasing behavior, the recommendation cycle can be appropriately extended and the recommended content can be fine-tuned; for users who are sensitive to recommendations, recommended products are updated in a timely manner according to their recent changes in interests.

[0046] S34, applying the formulated update strategy to the periodic recommendation database, updating product recommendation information, recommendation time plan and other contents in real time, and sending the optimized recommendation information of the updated periodic recommendation database to the corresponding registered users.

[0047] As described in the above steps S28-S34, the present invention can continuously obtain new actual purchase data of registered users and organize it into a structured format, which makes the monitoring of user purchase behavior dynamic and uninterrupted. It can grasp the latest consumption situation of users in real time, and provide a timely and detailed data basis for the subsequent accurate analysis of changes in user needs and evaluation of recommendation effects. Different from the practice of relying solely on periodic data, it ensures that the recommendation system can be flexibly adjusted according to the current actual situation of the user, always fit the user's changing purchase behavior, carefully compare the proposed recommended information with the actual purchase data, and clarify the differences between the two, which helps to accurately lock the specific points where the recommendation deviation occurs. Knowing the difference between the actual purchase of users and the recommended products clearly, you can specifically dig out the reasons behind it, such as whether the recommended product category does not match, or whether it is affected by factors such as price and brand, to find the right direction for subsequent in-depth analysis and improvement of recommendation strategies, so that the optimization recommendation is more targeted, and avoid blind adjustments that cannot solve actual problems.

[0048] It is also possible to formulate targeted update strategies based on the classification of user groups and the results of behavioral pattern recognition, which fully demonstrates the advantages of flexibility. According to the characteristics of different user groups, such as appropriately extending the recommendation cycle for those with stable purchasing behavior and timely updating recommended products for those who are sensitive to recommendations, the recommendation adjustment can be closely matched with the actual needs of each group, realizing personalized recommendation optimization of "thousands of faces for thousands of people", meeting the different needs of different users at different stages to the greatest extent, improving user experience, making recommendation information easier for users to recognize and adopt, applying the formulated update strategy to the periodic recommendation database in real time and sending optimized recommendation information, ensuring that the recommendation system can quickly respond to the analysis results and present the improved recommendations to users in a timely manner. This timeliness allows users to feel that the recommended content becomes more in line with their needs as soon as possible, enhances the positive interaction between users and the recommendation system, maintains users' favorability towards the recommendation service, and ensures that the recommendation always keeps up with the pace of user behavior and market changes, and continuously optimizes the recommendation effect.

[0049] The present invention also discloses a big data analysis and processing system for personalized recommendation, comprising: A first acquisition module, used to acquire user information of a registered user, wherein the user information includes preference feature data and taboo feature data; The second acquisition module is used to obtain the current popular product information of this platform; A first analysis module is used to compare and analyze the preference feature data with the current popular product information to obtain recommendation priority classification data; Establishing a module for combining and analyzing the recommendation priority classification data with the taboo feature data and establishing a periodic recommendation database; A third acquisition module is used to acquire the proposed recommendation information according to the periodic recommendation database; The fourth acquisition module is used to obtain the actual purchase data of the registered user; The second analysis module is used to update the periodic recommendation database in real time based on the comparison and analysis between the proposed recommendation information and the actual purchase data; The recommendation module is used to send optimized recommendation information to corresponding registered users according to the updated periodic recommendation database.

[0050] Specifically, the first acquisition module includes: A first acquisition unit is used to acquire the user's behavior data on social media by connecting the data interface with the data collection tool and in accordance with the data acquisition rules and privacy policy of this platform; A processing unit, used to convert and standardize the acquired behavior data to make it compatible with the data structure of this platform; A correction unit, used to correct the processed behavior data, remove erroneous data therein, and obtain preference feature data; The second acquisition unit is used to analyze the return and exchange reason records of users after purchase, extract information related to the product being unsuitable or not meeting expectations, and acquire taboo feature data.

[0051] The present application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0052] The present application also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above method when executed by a processor.

[0053] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, value library or other media provided in this application and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0054] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0055] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent results or equivalent process changes made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A big data analysis and processing method for personalized recommendation, characterized in that: include: Acquire user information of a registered user, wherein the user information includes preference feature data and taboo feature data; Get the current hot product information on this platform; Comparing and analyzing the preference feature data with the current popular product information, obtaining recommendation priority classification data; Performing combined analysis based on the recommendation priority classification data and the taboo feature data, and establishing a periodic recommendation database; Acquiring proposed recommendation information according to the periodic recommendation database; Get the actual purchase data of registered users; Based on the comparison and analysis between the proposed recommendation information and the actual purchase data, the periodic recommendation database is updated in real time; Send optimized recommendation information to corresponding registered users based on the updated periodic recommendation database.

2. The big data analysis and processing method for personalized recommendation according to claim 1, characterized in that: The steps to obtain the user information of a registered user include: By connecting the data interface with the data collection tool and following the data acquisition rules and privacy policy of this platform, we can obtain the user's behavior data on social media; Convert and standardize the acquired behavior data to make it compatible with the data structure of this platform; Correct the processed behavioral data, remove erroneous data, and obtain preference feature data; By analyzing the return and exchange reason records of users after purchase, we can extract information related to the product being unsuitable or not meeting expectations and obtain taboo feature data.

3. According to the big data analysis and processing method for personalized recommendation of claim 1, it is characterized in that: The step of obtaining recommendation priority classification data comprises: Create a two-dimensional matrix, where the rows represent the user's preference data features and the columns represent the popular product information; According to the matching degree between the user preference feature data and the popular product information attributes, the corresponding score is filled in the matrix. If the user's preferred product category completely matches the popular product category, the score is 10, if it partially matches, the score is 5; if it does not match, the score is 0; For each popular product, calculate its score in terms of preference matching, which is calculated as: ； in, is the preference matching score of the ith popular product, is the weight of the jth preference feature, is the score of user preference feature j and popular product i in the matching matrix, and n is the number of preference features. quantity; Taking into account the priority indicators of popular products, we set weights for each indicator, where the weight of sales volume is 0.4, the weight of sales revenue is 0.3, the weight of browsing popularity is 0.2, and the weight of collection volume is 0.1; The priority score calculation formula is: ； in, is the popularity score of the ith popular product, is the standardized value of sales volume of the ith product, is the normalized value of sales, is the normalized value of browsing popularity, is the standardized value of collection quantity; According to the priority level of different products Sort in descending order to obtain recommended priority grading data.

4. The big data analysis and processing method for personalized recommendation according to claim 1, characterized in that: The step of acquiring the proposed recommendation information according to the periodic recommendation database comprises: Collect and integrate user preference feature data, taboo feature data, recommendation priority classification data, and various product related information to build a complete data set; Based on the taboo feature data, check the products in the recommendation priority classification data one by one, eliminate products containing taboo features, and ensure that the recommended products meet the basic requirements of users; Analyze user behavior and product usage cycle patterns, and determine the recommended time node or time range in the cycle recommendation database for the screened products based on the current time, so as to achieve accurate time planning and establish a cycle recommendation database; Monitor the user's contextual information and recent behavior trends in real time, match them with the product information in the periodic recommendation database, and select products that match the user's current situation and interests; Based on the matching results, personalized customization is performed according to the user's device and usage scenario, and the selected product information is converted into recommended information suitable for display.

5. The big data analysis and processing method for personalized recommendation according to claim 1, characterized in that: The step of obtaining the actual purchase data of the registered user comprises: Obtain purchase records of registered users extracted from the transaction database, including details of purchased products or services, purchase time, purchase amount, and payment method; Filter out valid and complete purchase data, and exclude test data, error records, and incomplete transaction information; Clean the extracted purchase data, deal with noise and outliers in the data, correct obviously erroneous price data, and unify the date; Integrate purchase data with other relevant information of the user, associate data from different sources through the user's unique identifier, and build a complete user purchase behavior data set; Store the processed actual purchase data in the data analysis storage area or database table, establish a data update mechanism, and regularly obtain the latest purchase data.

6. The big data analysis and processing method for personalized recommendation according to claim 1, characterized in that: The step of sending the optimization recommendation information to the corresponding registered user according to the updated periodic recommendation database includes: Continuously obtain new actual purchase data of registered users, including purchase product information, purchase time, and purchase quantity, and organize them into a structured format; Extract the historical recommendation records of relevant products and the corresponding recommendation time and recommendation strategy information from the periodic recommendation database, and conduct a comprehensive comparative analysis; Compare the proposed recommendation information with the actual purchase data in detail to determine whether the product actually purchased by the user is within the proposed recommendation range, and the difference between the purchased product and the recommended product; Based on the user's purchase behavior, feedback data and calculated deviation values, the user is divided into different groups using clustering algorithms or classification models; Analyze the behavior patterns and preference trends of different user groups and identify common patterns that lead to recommendation bias; According to the user group classification and behavior pattern recognition results, formulate targeted periodic recommendation database update strategies for different groups; Apply the formulated update strategy to the periodic recommendation database, update the product recommendation information and recommended time plan content in real time, and send the optimized recommendation information of the updated periodic recommendation database to the corresponding registered users.

7. A big data analysis and processing system for personalized recommendation, characterized in that: include: A first acquisition module, used to acquire user information of a registered user, wherein the user information includes preference feature data and taboo feature data; The second acquisition module is used to obtain the current popular product information of this platform; A first analysis module is used to compare and analyze the preference feature data with the current popular product information to obtain recommendation priority classification data; Establishing a module for combining and analyzing the recommendation priority classification data with the taboo feature data and establishing a periodic recommendation database; A third acquisition module is used to acquire the proposed recommendation information according to the periodic recommendation database; The fourth acquisition module is used to obtain the actual purchase data of the registered user; The second analysis module is used to update the periodic recommendation database in real time based on the comparison and analysis between the proposed recommendation information and the actual purchase data; The recommendation module is used to send optimized recommendation information to corresponding registered users according to the updated periodic recommendation database.

8. A big data analysis and processing system for personalized recommendation according to claim 7, characterized in that: The first acquisition module includes: A first acquisition unit is used to acquire the user's behavior data on social media by connecting the data interface with the data collection tool and in accordance with the data acquisition rules and privacy policy of this platform; A processing unit, used to convert and standardize the acquired behavior data to make it compatible with the data structure of this platform; A correction unit, used to correct the processed behavior data, remove erroneous data therein, and obtain preference feature data; The second acquisition unit is used to analyze the return and exchange reason records of users after purchase, extract information related to the product being unsuitable or not meeting expectations, and acquire taboo feature data.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of a big data analysis and processing method for personalized recommendation described in any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a big data analysis and processing method for personalized recommendation described in any one of claims 1 to 6 are implemented.

Images