Intelligent order recommendation method and system

Abstract

The invention relates to the technical field of business management, in particular to an order intelligent recommendation method and system, and the method comprises the steps: data collection and integration, user portrait construction, recommendation algorithm application and data updating. Compared with the prior art that a simple matching recommendation scheme possibly adopted in the prior art neglects the comprehensiveness of commodity attributes and the complexity of user requirements, so that a recommendation result is not accurate and personalized enough, the scheme comprises the following steps: firstly, determining detailed commodity attributes including categories, prices, brands, evaluations and the like according to commodity information; then, an appropriate recommendation algorithm is carefully selected according to business requirements and data characteristics, and parameter optimization is carried out; inputting the user portrait and the commodity feature data into an algorithm for execution, generating a recommendation result, and performing fine screening; finally, the recommendation results are intelligently sorted according to the user portraits and the commodity attributes and output to the users, the recommendation accuracy and individuation degree are remarkably improved through the scheme, and the shopping experience and satisfaction degree of the users are effectively enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of business management, and particularly to an intelligent order recommendation method and system. Background Art

[0002] In the field of e-commerce, intelligent order recommendation systems and methods play a crucial role. They aim to provide personalized shopping suggestions for users by analyzing user behavior and product information, thereby enhancing the user experience and sales conversion rate. However, existing recommendation technologies face a series of challenges during implementation.

[0003] Traditional recommendation schemes often rely on simple matching logic or recommend based on only a small number of product attributes. These schemes lack comprehensiveness in dealing with product attributes. They may only consider basic information such as product category or price, while ignoring more dimensions of attributes such as brand, user reviews, and product descriptions. This simplified processing method is difficult to accurately capture the subtle differences between products, resulting in limited accuracy and personalization of the recommendation results.

[0004] At the same time, the selection of recommendation algorithms in existing technologies also often lacks pertinence. Different business scenarios and data characteristics require different recommendation algorithms to adapt, but traditional schemes often adopt a one-size-fits-all strategy and do not carefully select algorithms and optimize parameters according to actual needs. This approach limits the flexibility and accuracy of the recommendation system and is difficult to meet the personalized needs of different users.

[0005] In addition, there are also deficiencies in the processing and display of recommendation results in existing technologies. Traditional recommendation schemes often lack fine screening and intelligent sorting of recommendation results, resulting in the recommendation list that users see may contain a large number of products that do not match their interests and needs. This not only reduces the user's shopping experience but also may trigger a sense of distrust in the recommendation system. Summary of the Invention

[0006] In order to overcome the problems mentioned in the above background art, the present invention proposes an intelligent order recommendation method and system.

[0007] The technical solution of the present invention is: an intelligent order recommendation method, including the following steps: S11: Data collection and integration, collecting data related to orders, where the data related to orders includes user behavior data, product information data, and promotional activity data; S12: User portrait construction, constructing a basic user portrait based on the collected user behavior data to identify the user's interest preferences and purchase habits; S13: Application of the recommendation algorithm, using the set recommendation algorithm to recommend products for users according to the user's basic portrait; S14: Data update. Introduce a time decay factor to assign different impact weights to users' historical behavior data, and update and adjust the recommendation algorithm based on users' real-time behavior data and market trends.

[0008] Preferably, when collecting data related to orders, the following steps are included: S21: Data collection. First, determine the data sources and design a data collection plan, and then collect data according to the data collection plan. Among them, the data sources include internal data sources and external data sources; S22: Data processing. First, clean the data, including removing invalid data, handling missing values, and correcting incorrect data, and then convert the data format to convert data from different sources into a unified format; S23: Data integration. Integrate the data after cleaning and format conversion into a unified data warehouse, and perform data verification on the integrated data.

[0009] Preferably, when constructing a user's basic portrait based on the collected user behavior data to identify the user's interest preferences and purchase habits, the following steps are included: S31: Feature extraction. Use a set feature extraction algorithm to extract the required feature data from the data after cleaning and integration; S32: User clustering. Group users according to set common features; S33: User portrait construction. Based on the extracted features and clustering results, construct a detailed portrait for each user group.

[0010] Preferably, when grouping users according to set common features, the principle formula is: ; Wherein, represents the similarity between user x and y, and x i and y i are respectively the observed values of the i-th behavior data of user x and user y.

[0011] Preferably, when constructing a detailed portrait for each user group based on the extracted features and clustering results, first decompose the user's behavior data. Among them, the principle formula for decomposing the user's behavior data is: User behavior data = time + location + person + action + action object + tool + metric; Wherein, the metric is a set measurement method, including price and quantity; And the principle expression formula of the user portrait is: User portrait = explicit features + implicit features = physical features + psychological activities + behavioral features; Among them, physical characteristics include age, gender, and region, and behavioral characteristics include interests, preferences, and purchase habits.

[0012] Preferably, when using a set recommendation algorithm to recommend products to users based on the user's basic profile, the following steps are included: S41: Determining product attributes. According to product information, obtain the specific attributes of the product. Among them, the specific attributes of the product include the category, price, brand, and evaluation of the product; S42: Setting the recommendation algorithm. Select a suitable recommendation algorithm according to business requirements and data characteristics, and set parameters for the selected recommendation algorithm; S43: Executing the algorithm. Input the user profile and product feature data into the recommendation algorithm, execute the algorithm to generate a recommendation result, and screen the generated recommendation result to remove products that do not meet the user's needs; S44: Sorting and output. Sort the recommendation result according to the user's profile and the specific attributes of the product, and output it to the user.

[0013] Preferably, when obtaining the specific attributes of the product according to the product information, the following steps are included: S51: Analyzing the user profile. Collect and organize the basic information of product purchasers, analyze the behavioral data of purchasers, and evaluate the psychological characteristics of purchasers; S52: Determining key attributes. According to the results of user profile analysis, identify the product attributes that are most relevant to the characteristics and preferences of purchasers; S53: Refining and quantifying attributes. Refine and quantify each key attribute; S54: Verifying and outputting attributes. By comparing with actual purchase data, verify whether the selected attributes accurately reflect the characteristics and preferences of purchasers, obtain the specific attributes of the product, and output them.

[0014] Preferably, when selecting a suitable recommendation algorithm according to business requirements and data characteristics and setting parameters for the selected recommendation algorithm, the selected recommendation algorithms include: A11: Content-based recommendation algorithm. Recommend similar items based on the user's past browsing records and behaviors; A12: Collaborative filtering recommendation algorithm. Form recommendations by analyzing similar users and similar items in the user group; A13: Association rule recommendation algorithm. Based on association rule mining technology, discover the relevance of different items during the sales process, and thus make recommendations; A14: Utility-based recommendation algorithm. A utility function is created for each user. The utility function takes into account the user's preferences for various attributes of items. Then, based on the utility function, the expected utility of the user for unexposed items is calculated, and the item with the highest expected utility is selected for recommendation.

[0015] Preferably, when inputting user portraits and product feature data into the recommendation algorithm, executing the algorithm to generate recommendation results, and screening the generated recommendation results to remove products that do not meet user needs, the principle formula of the recommendation algorithm is: ; Among them, sim(A, B) represents the matching degree between the user portrait and the product feature data. A represents the user portrait, and B represents the product feature data. and respectively represent the norms of A and B.

[0016] An intelligent order recommendation system includes: A data collection and integration module, responsible for collecting data related to orders, including user behavior data, product information data, and promotion activity data; A user portrait construction module, used to construct the user's basic portrait based on the collected user behavior data; A recommendation algorithm application module, used to select a suitable recommendation algorithm according to business requirements and data characteristics, and input the user portrait and product feature data into the recommendation algorithm to generate recommendation results; A data update and maintenance module, used to introduce a time decay factor to assign different influence weights to the user's historical behavior data; A product attribute determination module, responsible for collecting product information and determining the specific attributes of the product according to the product information; A user interface and interaction module, responsible for interacting with the user and displaying the recommendation results; A log and monitoring module, used to record the running log of the system and monitor the performance and stability of the system.

[0017] The beneficial effects of the present invention: 1. Compared with the simple matching or recommendation schemes based on a small number of product attributes that may be adopted in the prior art, these schemes often neglect the comprehensiveness of product attributes and the complexity of user needs, resulting in inaccurate and non-personalized recommendation results. This scheme, however, adopts a more comprehensive and refined recommendation process: First, determine detailed product attributes based on product information, including category, price, brand, and evaluation, etc.; then carefully select a suitable recommendation algorithm according to business requirements and data characteristics and optimize the parameters; then input user portrait and product feature data into the algorithm to execute, generate recommendation results and conduct fine screening; finally, intelligently sort the recommendation results according to user portrait and product attributes and output them to users. This scheme significantly improves the accuracy and personalization of recommendations, effectively enhancing the user's shopping experience and satisfaction; 2. Compared with the scheme in the prior art that may extract attributes only based on basic product information, this scheme often ignores the important influence of the characteristics and preferences of the purchasers on the selection of product attributes, resulting in inaccurate attribute selection and difficulty in effectively meeting user needs. This scheme, however, accurately identifies the key product attributes most relevant to the characteristics and preferences of the purchasers by deeply analyzing the user portrait, including the basic information, behavior data, and psychological characteristics of the purchasers, and further refines and quantifies these attributes. Finally, through comparison and verification with actual purchase data, it is ensured that the selected attributes can accurately reflect the characteristics and preferences of the purchasers, thereby obtaining more accurate and personalized specific product attributes. This scheme significantly improves the accuracy and pertinence of product attribute extraction, provides more reliable data support for subsequent personalized recommendations, and effectively enhances the accuracy of recommendations and user satisfaction; 3. Compared with the drawback in the prior art that may construct user portraits only relying on limited data dimensions, resulting in insufficiently refined portraits that are difficult to accurately reflect users' interest preferences and purchase habits, this scheme uses a feature extraction algorithm to deeply mine the required features from multi-source data that has been strictly cleaned and integrated, and then scientifically groups users according to common features through a refined user clustering strategy. Finally, based on these detailed features and clustering results, a more accurate and comprehensive portrait is constructed for each user group. This scheme significantly improves the richness and accuracy of user portraits, provides strong support for subsequent personalized recommendations, and effectively enhances the user experience and satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 The flowchart showing the order intelligent recommendation method of the present invention is presented; Figure 2 The structural diagram showing the order intelligent recommendation system of the present invention is presented. DETAILED DESCRIPTION OF THE INVENTION

[0019] The present invention will be further described below in conjunction with the drawings and embodiments.

[0020] Please refer to Figure 1 , the present invention provides an embodiment: an order intelligent recommendation method, including the following steps: S11: Data collection and integration, collecting data related to orders, wherein the data related to orders includes user behavior data, commodity information data, and promotion activity data; S12: User profile construction, constructing a basic user profile based on the collected user behavior data, and identifying the user's interest preferences and purchase habits; S13: Recommendation algorithm application, using the set recommendation algorithm, and recommending commodities to users according to the basic user profile; S14: Data update, introducing a time decay factor, assigning different influence weights to the user's historical behavior data, and updating and adjusting the recommendation algorithm according to the user's real-time behavior data and market trends.

[0021] As described above, the present invention comprehensively collects and integrates order-related data, accurately constructs a user profile to identify their interest preferences and purchase habits, and then applies an efficient recommendation algorithm to achieve personalized commodity recommendations. At the same time, this solution also pays attention to the real-time update of data, introduces a time decay factor to optimize the influence weight of historical behavior data, and closely follows the market trends and the user's real-time behavior to adjust the recommendation algorithm, thereby ensuring the accuracy and timeliness of the recommendation results, and greatly improving the user experience and order conversion rate.

[0022] Preferably, when collecting data related to orders, it includes the following steps: S21: Data collection, first determining the data sources and designing a data collection plan, and then collecting data according to the data collection plan, wherein the data sources include internal data sources and external data sources; S22: Data processing, first cleaning the data, including removing invalid data, processing missing values, and correcting incorrect data, and then performing format conversion on the data to convert data from different sources into a unified format; S23: Data integration, integrating the data after cleaning and format conversion into a unified data warehouse, and performing data verification on the integrated data.

[0023] As described above, the present invention ensures the comprehensiveness and accuracy of the data by clarifying the data sources and carefully designing the collection plan. Through strict data cleaning and format conversion processes, invalid and incorrect data are effectively removed, the data format is unified, and the data quality is improved. Finally, the data is integrated into a unified data warehouse and verified, providing a solid data foundation for subsequent order intelligent recommendations, and further enhancing the reliability and effectiveness of the recommendation system.

[0024] Preferably, when constructing a user's basic profile based on the collected user behavior data to identify the user's interest preferences and purchase habits, the following steps are included: S31: Feature extraction, using a set feature extraction algorithm to extract the required feature data from the data that has been cleaned and integrated; S32: User clustering, grouping users according to the set common features; S33: User profile construction, constructing a detailed profile for each user group based on the extracted features and clustering results.

[0025] As described above, compared with the disadvantage in the prior art that the construction of user profiles may only rely on limited data dimensions, resulting in insufficiently detailed profiles that are difficult to accurately reflect users' interest preferences and purchase habits, this solution uses a feature extraction algorithm to deeply mine the required features from multi-source data that has been strictly cleaned and integrated. Then, through a refined user clustering strategy, users are scientifically grouped according to common features. Finally, based on these detailed features and clustering results, a more accurate and comprehensive profile is constructed for each user group. This solution significantly improves the richness and accuracy of user profiles, provides strong support for subsequent personalized recommendations, and effectively enhances the user experience and satisfaction.

[0026] Preferably, when grouping users according to the set common features, the principle formula is: ; where, represents the similarity between user x and y, and x i and y i are respectively the observed values of the i-th behavior data of user x and user y.

[0027] As described above, the present invention accurately measures the similarity of behavior data between users by using a specific similarity calculation formula. This formula comprehensively considers the observed values of users on different behavior data, so as to scientifically group users with similar interest preferences and purchase habits into the same group. This refined user clustering strategy not only improves the accuracy of user profiles, but also provides more reliable data support for subsequent personalized recommendations, significantly enhancing the effect of the recommendation system and the user experience.

[0028] Preferably, when constructing a detailed profile for each user group based on the extracted features and clustering results, first decompose the user's behavior data. Among them, the principle formula for decomposing the user's behavior data is: User behavior data = time + location + person + action + action object + tool + metric; where, the metric is a set measurement method, including price and quantity; And the principle expression formula of the user portrait is as follows: User portrait = explicit features + implicit features = physical features + psychological activities + behavioral features; Among them, physical features include age, gender, and region, and behavioral features include interests, preferences, and purchase habits.

[0029] As described above, the present invention first uses a refined principle formula for decomposing behavioral data to carefully split the user's behavioral data according to a set measurement method (such as price and quantity), so as to deeply explore the internal laws and characteristics of user behavior. At the same time, combined with the principle expression formula of the user portrait, comprehensively considering the user's physical features (such as age, gender, region) and behavioral features (such as interests, preferences, purchase habits), a comprehensive and accurate portrait is constructed for each user group. This technical solution not only improves the richness and depth of the user portrait, but also provides more detailed data support for subsequent personalized recommendations, effectively enhancing the pertinence of the recommendation system and user satisfaction.

[0030] Preferably, when using a set recommendation algorithm to recommend products to users based on the user's basic portrait, the following steps are included: S41: Determine product attributes. According to the product information, obtain the specific attributes of the product. Among them, the specific attributes of the product include the category, price, brand, and evaluation of the product; S42: Set the recommendation algorithm. Select a suitable recommendation algorithm according to business requirements and data characteristics, and set parameters for the selected recommendation algorithm; S43: Execute the algorithm. Input the user portrait and product feature data into the recommendation algorithm, execute the algorithm to generate a recommendation result, and screen the generated recommendation result to remove products that do not meet the user's needs; S44: Sort and output. Sort the recommendation result according to the user's portrait and the specific attributes of the product, and output it to the user.

[0031] As described above, compared with the simple matching or recommendation schemes based on a small number of product attributes that may be adopted in the prior art, these schemes often ignore the comprehensiveness of product attributes and the complexity of user needs, resulting in inaccurate and non-personalized recommendation results. This scheme adopts a more comprehensive and refined recommendation process: first, determine detailed product attributes based on product information, including category, price, brand, and evaluation, etc.; then carefully select a suitable recommendation algorithm according to business requirements and data characteristics and optimize the parameters; then input the user profile and product feature data into the algorithm for execution, generate recommendation results and conduct fine screening; finally, perform intelligent sorting on the recommendation results according to the user profile and product attributes and output them to the user. This scheme significantly improves the accuracy and personalization of recommendations, effectively enhancing the user's shopping experience and satisfaction.

[0032] Preferably, when obtaining the specific attributes of a product according to product information, the following steps are included: S51: User profile analysis, collect and organize the basic information of product purchasers, analyze the behavior data of purchasers, and evaluate the psychological characteristics of purchasers; S52: Determine key attributes, according to the results of user profile analysis, identify the product attributes most relevant to the characteristics and preferences of purchasers; S53: Attribute refinement and quantification, refine and quantify each key attribute; S54: Attribute verification and output, by comparing with actual purchase data, verify whether the selected attributes accurately reflect the characteristics and preferences of purchasers, obtain the specific attributes of the product and output them.

[0033] As described above, compared with the scheme that may extract attributes only based on the basic information of products in the prior art, this scheme often ignores the important influence of the characteristics and preferences of purchasers on the selection of product attributes, resulting in inaccurate attribute selection and difficulty in effectively meeting user needs. This scheme accurately identifies the key product attributes most relevant to the characteristics and preferences of purchasers by deeply analyzing the user profile, including the basic information, behavior data, and psychological characteristics of purchasers, and further refines and quantifies these attributes. Finally, through comparison and verification with actual purchase data, it is ensured that the selected attributes can accurately reflect the characteristics and preferences of purchasers, so as to obtain more accurate and personalized specific product attributes. This scheme significantly improves the accuracy and pertinence of product attribute extraction, provides more reliable data support for subsequent personalized recommendations, and effectively enhances the accuracy of recommendations and user satisfaction.

[0034] Preferably, when selecting a suitable recommendation algorithm according to business requirements and data characteristics and setting parameters for the selected recommendation algorithm, the selected recommendation algorithms include: A11: Content-based recommendation algorithm, which recommends similar items based on the user's past browsing records and behaviors; A12: Collaborative filtering recommendation algorithm, which forms recommendations by analyzing similar users and similar items in the user group; A13: Association rule recommendation algorithm, which discovers the correlation between different items during the sales process based on association rule mining technology and then makes recommendations; A14: Utility-based recommendation algorithm, which creates a utility function for each user. The utility function takes into account the user's preferences for various attributes of the item, then calculates the expected utility of the user for items not yet encountered based on the utility function, and selects the item with the highest expected utility for recommendation.

[0035] As described above, when the present invention selects a recommendation algorithm according to business requirements and data characteristics, this technical solution comprehensively considers a variety of advanced recommendation algorithms, including the content-based recommendation algorithm, which can accurately match the user's historical interests; the collaborative filtering recommendation algorithm, which effectively mines the similarity in the user group to improve the relevance of recommendations; the association rule recommendation algorithm, which deeply mines the sales associations between items to broaden the recommendation horizon; and the utility-based recommendation algorithm, which accurately calculates the user's expected utility through a personalized utility function to ensure that the recommendations highly conform to the user's needs. This diversified algorithm selection strategy, combined with fine parameter settings, not only improves the diversity and accuracy of recommendations, but also significantly enhances the flexibility and adaptability of the recommendation system, and can better meet the personalized needs of different users.

[0036] Preferably, when inputting the user profile and product feature data into the recommendation algorithm, executing the algorithm to generate a recommendation result, and screening the generated recommendation result to remove products that do not meet the user's needs, the principle formula of the recommendation algorithm is: ; Among them, sim(A,B) represents the matching degree between the user profile and the product feature data, A represents the user profile, and B represents the product feature data, and respectively represent the norms of A and B.

[0037] As described above, the present invention combines user portraits with product feature data through an advanced recommendation algorithm. This algorithm accurately calculates the degree of fit between the two based on the matching formula between the user portrait (A) and the product feature data (B), where the matching degree is jointly determined by the norms of the features of both sides and their mutual relationship. This technical solution not only significantly improves the accuracy of recommendations, ensuring that the recommended products closely meet the personalized needs of users, but also, after the generation of the recommendation results, further executes a strict screening process to eliminate any products that do not meet user expectations, thus greatly optimizing the user experience and enabling users to efficiently and conveniently discover and obtain the products they are truly interested in.

[0038] Please refer to Figure 2 , the present invention provides an embodiment: an order intelligent recommendation system, including: A data collection and integration module, responsible for collecting data related to orders, including user behavior data, product information data, and promotion activity data; A user portrait construction module, used to construct a basic portrait of the user based on the collected user behavior data; A recommendation algorithm application module, used to select a suitable recommendation algorithm according to business requirements and data characteristics, and input the user portrait and product feature data into the recommendation algorithm to generate recommendation results; A data update and maintenance module, used to introduce a time decay factor to assign different influence weights to the user's historical behavior data; A product attribute determination module, responsible for collecting product information and determining the specific attributes of the product according to the product information; A user interface and interaction module, responsible for interacting with users and displaying the recommendation results; A log and monitoring module, used to record the running logs of the system and monitor the performance and stability of the system.

[0039] As described above, the present invention integrates multiple modules such as data collection and integration, user portrait construction, recommendation algorithm application, data update and maintenance, product attribute determination, user interface and interaction, and log and monitoring, forming a comprehensive and efficient recommendation system. Through accurate data collection and integration, the system can construct a detailed user portrait, select the most suitable recommendation algorithm based on business requirements and data characteristics, and generate highly personalized recommendation results. At the same time, the system also pays attention to the real-time update and maintenance of data to ensure the timeliness and accuracy of the recommendation results. In addition, through an intuitive user interface and interaction design, as well as a comprehensive log and monitoring function, the system not only improves the user experience, but also ensures the stable operation and continuous optimization of the system. This comprehensive technical solution significantly enhances the performance and effect of the recommendation system, providing users with a more intelligent and convenient shopping experience.

[0040] The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings. However, the present invention is not limited to the above embodiments, and various changes can be made without departing from the spirit of the present invention within the scope of knowledge possessed by those skilled in the art.

Claims

1. An intelligent order recommendation method, characterized in that: It includes the following steps: S11: Data collection and integration. Collect data related to orders, where the data related to orders includes users' behavioral data, product information data, and promotional activity data; S12: User portrait construction. Based on the collected users' behavioral data, construct the basic portrait of users, and identify users' interest preferences and purchase habits; S13: Recommendation algorithm application. Use the set recommendation algorithm to recommend products to users according to the basic portrait of users; S14: Data update. Introduce a time decay factor, assign different influence weights to users' historical behavioral data, and update and adjust the recommendation algorithm according to users' real-time behavioral data and market trends.

2. The intelligent order recommendation method according to claim 1, wherein: When collecting data related to orders, it includes the following steps: S21: Data collection. First, determine the data sources and design a data collection plan, and then collect data according to the data collection plan. Among them, the data sources include internal data sources and external data sources; S22: Data processing. First, clean the data, including removing invalid data, handling missing values, and correcting incorrect data, and then convert the format of the data to convert data from different sources into a unified format; S23: Data integration. Integrate the data after cleaning and format conversion into a unified data warehouse, and perform data verification on the integrated data.

3. The order intelligent recommendation method according to claim 2, characterized in that: When constructing the basic portrait of users based on the collected users' behavioral data and identifying users' interest preferences and purchase habits, it includes the following steps: S31: Feature extraction. Use the set feature extraction algorithm to extract the required feature data from the data after cleaning and integration; S32: User clustering. Group users according to the set common features; S33: User portrait construction. Based on the extracted features and clustering results, construct a detailed portrait for each user group.

4. An order intelligent recommendation method according to claim 3, characterized in that: When grouping users according to the set common features, the principle formula is: ； Among them, represents the similarity between user x and y, where x i and y i are respectively the observed values of the i-th behavior data of user x and user y.

5. The order intelligent recommendation method according to claim 4, characterized in that: When constructing a detailed portrait for each user group based on the extracted features and clustering results, first decompose the users' behavioral data. Among them, the principle formula for decomposing the users' behavioral data is: User behavioral data = time + location + person + action + action object + tool + metric; Among them, the metric is the set measurement method, including price and quantity; And the principle expression formula of the user portrait is: User portrait = explicit features + implicit features = physical features + psychological activities + behavioral features; Among them, the physical features include age, gender, and region, and the behavioral features include interests, preferences, and purchase habits.

6. The order intelligent recommendation method according to claim 5, wherein: When using the set recommendation algorithm to recommend products to users according to the basic portrait of users, it includes the following steps: S41: Product attribute determination. According to the product information, obtain the specific attributes of the product. Among them, the specific attributes of the product include the category, price, brand, and evaluation of the product; S42: Recommendation algorithm setting. Select a suitable recommendation algorithm according to business requirements and data characteristics, and set parameters for the selected recommendation algorithm; S43: Algorithm execution. Input the user profile and product feature data into the recommendation algorithm, execute the algorithm to generate recommendation results, and screen the generated recommendation results to remove products that do not meet the user's needs. S44: Sorting and output. Sort the recommendation results according to the user profile and the specific attributes of the products, and output them to the user.

7. An order intelligent recommendation method according to claim 6, characterized in that: When obtaining the specific attributes of a product according to the product information, the following steps are included: S51: User profile analysis. Collect and organize the basic information of product purchasers, analyze the behavior data of the purchasers, and evaluate the psychological characteristics of the purchasers. S52: Determine key attributes. According to the results of the user profile analysis, identify the product attributes that are most relevant to the characteristics and preferences of the purchasers. S53: Attribute refinement and quantification. Refine and quantify each key attribute. S54: Attribute verification and output. By comparing with the actual purchase data, verify whether the selected attributes accurately reflect the characteristics and preferences of the purchasers, and obtain and output the specific attributes of the products.

8. The order intelligent recommendation method according to claim 7, characterized in that: When selecting a suitable recommendation algorithm according to the business requirements and data characteristics and setting parameters for the selected recommendation algorithm, the selected recommendation algorithms include: A11: Content-based recommendation algorithm. Recommend similar items based on the user's past browsing records and behaviors. A12: Collaborative filtering recommendation algorithm. Form recommendations by analyzing similar users and similar items in the user group. A13: Association rule recommendation algorithm. Based on association rule mining technology, discover the relevance of different items during the sales process and make recommendations accordingly. A14: Utility-based recommendation algorithm. Create a utility function for each user. The utility function takes into account the user's preferences for various attributes of the items, then calculates the expected utility of the user for items not yet contacted according to the utility function, and selects the item with the highest expected utility for recommendation.

9. The order intelligent recommendation method according to claim 8, wherein: When inputting the user profile and product feature data into the recommendation algorithm, executing the algorithm to generate recommendation results, and screening the generated recommendation results to remove products that do not meet the user's needs, the principle formula of the recommendation algorithm is: ； Among them, sim(A, B) represents the matching degree between the user portrait and the commodity feature data, A represents the user portrait, and B represents the commodity feature data. and respectively represent the norms of A and B.

10. An order intelligent recommendation system, applied to an order intelligent recommendation method according to any one of claims 1-9, characterized in that: Including: Data collection and integration module. Responsible for collecting data related to orders, including user behavior data, product information data, and promotion activity data. User profile construction module. Used to construct the basic user profile based on the collected user behavior data. Recommendation algorithm application module. Used to select a suitable recommendation algorithm according to the business requirements and data characteristics, and input the user profile and product feature data into the recommendation algorithm to generate recommendation results. Data update and maintenance module. Used to introduce a time decay factor to assign different influence weights to the user's historical behavior data. Product attribute determination module. Responsible for collecting product information and determining the specific attributes of the products according to the product information. User interface and interaction module. Responsible for interacting with the user and displaying the recommendation results. Log and monitoring module. Used to record the running logs of the system and monitor the performance and stability of the system.

Images