E-commerce platform search recommendation and intelligent secretary integrated system

Abstract

The invention belongs to the technical field of e-commerce, and particularly relates to an e-commerce platform search recommendation and intelligent secretary integrated system, which comprises the following steps of: establishing an accurate user portrait by analyzing behavior data of a user and personalized preference of the user; through a basic search function, a commodity search function, a keyword matching function and an advanced function, a user is helped to quickly find a target commodity; personalized commodity recommendation is provided for the user based on collaborative filtering, and the user experience is improved; a virtual secretary function is provided, and a user is supported to perform natural language interaction; the behavior of the user on the platform is monitored and analyzed in real time, the demand and trend of the user are extracted, and searching and recommendation strategies are optimized; the system supports intelligent interaction of various input modes of voice, texts and pictures, searches commodities through voice or uploads pictures to search pictures through pictures, brings convenient, fast and personalized service experience for vast consumers, and also provides intelligent and efficient operation management for merchants.

Description

Technical Field

[0001] The present invention relates to the technical field of e-commerce, and in particular to an e-commerce platform search recommendation and intelligent secretary integrated system. Background Art

[0002] Existing e-commerce platforms generally use separate search and recommendation systems. The search system mainly relies on keyword matching and rule engines, which has limitations in accuracy and is difficult to meet users' diverse search needs. The recommendation system is mainly based on user historical behavior and item features, and there is a problem that the recommendation results lack personalization and guidance. At present, there is a lack of effective collaboration between the two functional modules of search and recommendation, and it is impossible to fully explore user intentions and product associations, resulting in poor user search experience and low product exposure.

[0003] In recent years, with the rapid development of artificial intelligence technology, smart secretary technology has gradually become a hot word in all walks of life. Smart secretaries have shown great potential in improving user experience and efficiency by understanding user natural language, predicting user needs and providing personalized services. At present, many smart assistant applications can provide some basic e-commerce services, such as product search and price comparison, but have not yet formed a complete integrated search and recommendation solution.

[0004] E-commerce platforms are facing increasingly fierce competition, and users' needs for product search and recommendation are becoming increasingly complex. Therefore, it is urgent to develop a system that can effectively combine search and recommendation functions and has intelligent interactive features to improve user shopping experience and merchant marketing efficiency. The system needs to be able to understand users' complex needs, accurately locate products, and provide personalized recommendations, so as to achieve seamless connection between search and recommendation, and provide users with convenient, efficient, and personalized e-commerce services like an intelligent secretary;

[0005] In view of the above technical defects, an integrated system solution of e-commerce platform search recommendation and intelligent secretary is proposed. Summary of the invention

[0006] To solve the above problems, the present invention provides the following technical solutions:

[0007] An e-commerce platform search recommendation and intelligent secretary integrated system, comprising:

[0008] The user portrait module builds accurate user portraits by analyzing user behavior data and personalized preferences, providing a basis for subsequent recommendations and search optimization;

[0009] Search engine module, including basic search functions, product search, keyword matching and advanced functions, fuzzy search, natural language understanding, to help users quickly find target products;

[0010] The recommendation system module provides personalized product recommendations to users based on collaborative filtering, content recommendation, and hybrid recommendation technologies, improves user experience, and makes accurate recommendations based on user historical behavior, popular products, and seasonal information;

[0011] The intelligent assistant module provides a virtual secretary function, supports users to interact with products in natural language, and can query products, set reminders, and manage shopping lists through voice or text input to improve interaction efficiency.

[0012] Data analysis module, which monitors and analyzes user behavior on the platform in real time, extracts user needs and trends, and optimizes search and recommendation strategies;

[0013] The multimodal interaction module supports intelligent interaction with multiple input methods such as voice, text, and pictures. You can search for products by voice or upload pictures for image search.

[0014] Furthermore, the user portrait module includes collecting user behavior data, browsing history, search keywords, click history, shopping cart items, purchase history, favorites, evaluations and feedback, obtaining basic information about users, collecting data from user login behavior, social media associated accounts, and other third-party platforms, analyzing user browsing, clicking, shopping frequency and behavior patterns, judging user interests and preferences for specific categories of goods, and discovering user active time, shopping cycle and user consumption habits during holidays and promotional periods through analysis of time periods. Based on user historical data, user interest tags are generated, and interest weights are dynamically adjusted based on user click-through rates and dwell time for different types of goods, providing priority references for the recommendation system, clustering users with similar behavior patterns to form specific user groups, and grouping users to accept different types of recommendation strategies to improve the accuracy of recommendations. Natural language processing technology is used to analyze user evaluations, feedback or customer service chat records to understand user emotional tendencies, provide a basis for the platform to improve products and services, and timely push relevant after-sales services or problem solutions. As user behavior changes, the portrait will be updated in real time, and new purchase behaviors, evaluations, and browsing behaviors will be immediately incorporated into the portrait model to ensure that the system responds to user interests in a timely manner.

[0015] Furthermore, the search engine module includes returning relevant product results based on the keywords input by the user, performing word segmentation processing on the natural language input by the user, establishing a search index, ensuring fast query and response, providing automatic correction or suggestions when the keywords input by the user are misspelled, calculating the relevance between the product and the search term based on the matching degree between the keyword and the product description, and sorting the search results based on the relevance, customizing the sorting of search results for each user in combination with user portraits, comprehensively considering multiple factors such as price, sales volume, evaluation, inventory, and brand, and sorting and optimizing the search results in multiple dimensions, understanding the user's complex queries through natural language processing technology, and for imprecise or vague search terms, the system provides similar or related products. When the user enters a search term, the system automatically associates related words based on the existing search history and popular products to help the user quickly find the desired product, and quickly filter products by clicking on different categories based on the product classification information, supporting refined filtering by color, size, and material product attributes to help users quickly locate specific products, and recommending other products or product combinations related to the search in combination with the user's search keywords and behavior history.

[0016] Furthermore, the search engine module also includes returning a list of products based on precise matching of keywords to ensure that core search needs are met, and increasing the recall range through a variety of expansion methods, such as synonym expansion, semantic analysis, and user profiling, to ensure that users can find relevant products even in the case of inaccurate searches. Through user uploads of pictures, the search engine returns similar or identical products based on image recognition technology, adjusts search results based on the user's geographic location, current activity, and time context information, and customizes the display of recommended products related to the current search term based on the user's historical search and purchase records. The validity of search results is identified through user click, purchase, and dwell time behavior data, and the search sorting strategy is continuously adjusted through these feedbacks. Through experiments on different search algorithms and sorting rules, the user search experience is optimized to ensure that the presented products are more in line with user needs, and to ensure that the inventory, price, and promotional information of the products are synchronized to the search results in real time to prevent users from purchasing sold out or expired products.

[0017] Furthermore, the recommendation system module includes collecting various behavioral data of users on the platform, including browsing records, search records, click histories, purchase records, and evaluations, cleaning, deduplicating, and normalizing the original data to ensure the quality and consistency of the data for subsequent model training and application of recommendation algorithms, based on user collaborative filtering, and according to the preferences of similar users, recommending the current user the products that the user likes. Based on the similarity between the products, recommending similar products to the user, based on product attributes and user preferences, recommending products similar to the products that the user has browsed or purchased in the past, comprehensively using collaborative filtering, content recommendation, and rule recommendation methods, dynamically adjusting recommendation strategies according to different scenarios and user characteristics, using neural network models to process complex user behavior data, and mining deep user interest relationships to generate more accurate recommendations, closely integrating the recommendation system with the user portrait module, providing highly personalized recommendations based on the user's age, gender, geographic location, and interest tag information, and adjusting the recommended content in real time according to the user's current behavior and contextual information.

[0018] Furthermore, the intelligent assistant module supports users to interact with the assistant through voice or text through natural language processing and understanding, understands the user's real needs through semantic analysis and intent recognition, supports multiple rounds of dialogue between users and assistants, and continuously maintains context information in complex problem situations. Based on the keywords entered by the user, the intelligent assistant can quickly help users search for related products and recommend products that meet the needs. Combined with user portraits, the intelligent assistant recommends products of interest to users during the interaction process. When a user asks about a product, the intelligent assistant provides detailed product descriptions, including specifications, user reviews, and price fluctuation information, provides users with current discount information and coupons, and predicts price changes of products to help users make purchasing decisions, help users quickly add products to shopping carts, view shopping cart contents, remind users to use coupons, and even guide users to complete settlement. The order status can be checked at any time through the intelligent assistant, including delivery progress and estimated delivery time. The intelligent assistant assists users in handling returns, exchanges, and complaints about after-sales issues, and guides users to complete operations according to the type of problem, providing users with automatic answers to common problems.

[0019] Furthermore, the data analysis module includes collecting data from different sources, including users' browsing behaviors, search histories, shopping records, evaluation data, and logistics information, supporting data integration from multiple channels such as websites, Apps, and mini-programs, processing different formats of data, including users' order records, user comments, and customer service conversation data, supporting the collection and processing of real-time data, including users' current browsing, clicking, and searching behaviors, facilitating timely adjustment of recommendation or promotion strategies, removing duplicate, abnormal, or missing data, ensuring the accuracy and consistency of data analysis, annotating and classifying the collected data, and normalizing the data, including unifying the time format and standardizing the values, to ensure the mutual comparison and analysis of various types of data. Based on users' browsing, searching, and purchasing behaviors, analyze users' interests and preferences to form personalized user portraits to support precise recommendations. Classify users into different groups according to their behavioral characteristics for targeted marketing and promotion, identify users at risk of churn, predict which users will stop using the platform through behavioral patterns, and provide corresponding retention strategies, such as targeted discounts or reminder notifications. Track the sales situation of different products, including sales volume, conversion rate, and inventory status, to help merchants optimize the product portfolio and replenishment strategies. Identify the advantages and disadvantages of products by analyzing users' evaluations and feedback on products to help merchants improve product quality and optimize marketing strategies. Track changes in product prices, analyze price fluctuations during different time periods or promotional periods, as well as the pricing strategies of competitors, and predict future sales trends based on historical sales data and market dynamics to help the platform manage product inventory and make market predictions. Evaluate the effects of various promotional activities, understand the impact of different promotional methods on sales volume, conversion rate, and user engagement, analyze the performance of different sales channels, and optimize resource allocation. By analyzing users' browsing paths on the platform, understand the pages where users stay, bounce rate, and key nodes in the conversion funnel during the access process to help optimize the user experience. Display the click distribution of users on the page through heatmaps to identify the key areas of user attention and help the platform optimize page design and content layout. Analyze the entire process from user browsing to final order placement, identify the links with low conversion rates, and put forward optimization suggestions.

[0020] Furthermore, the data analysis includes analyzing the data in the database, including visual analysis of building data correlation and similarity functions through function inner product, as follows:

[0021]

[0022] Among them, <f, g> is the function inner product, f(x) is the similarity function of the data in the database, g(x) is the correlation function of the data in the database, a and b are real numbers representing the lengths of the intercepted data segments in the database. Perform the operation of function inner product on the data according to the lengths of the data in the database, centrally process the similar and relevant data of the data, and conduct visual analysis.

[0023] Furthermore, the data analysis includes denoising the data in the database, including:

[0024]

[0025] Among them, λ is the universal threshold, σ is the database standard deviation, and N is the signal length.

[0026] Furthermore, the multimodal interaction module includes multiple languages ​​input through voice or text, and provides corresponding responses to meet international needs. Through NLP technology, it understands the user's true intentions, supports multiple rounds of dialogue, searches by uploading pictures, identifies products based on image recognition technology and provides relevant recommendations, recognizes user gestures, expressions, and objects through the camera, and provides corresponding feedback based on the image content, recognizes user gesture commands through the camera, and users control the interface, slide products, and zoom in and out pictures through gestures. It supports interaction with the system through screen touch, is suitable for mobile devices and touch screen devices, and inputs questions through keyboard or touch. The system uses text analysis technology to understand user needs and provide accurate replies when a question or command is given. It presents replies in multiple forms, including text, pictures, and links, to provide a richer interactive experience. It combines information from multiple modalities, including voice, text, and images, to conduct comprehensive analysis and response, to ensure that the interaction process of different modalities is smooth and natural, and users can switch seamlessly between voice, image, and text modalities. It analyzes users' emotions and feelings through voice, text, and images, adjusts the interaction method according to the user's emotional state, and provides personalized recommendations based on the user's different input modalities. It supports users to query products or information through different input methods and conduct accurate searches in combination with other modalities.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] The present invention discloses an e-commerce platform search recommendation and intelligent secretary integrated system. By analyzing the user's behavior data and the user's personalized preferences, an accurate user portrait is established to provide a basis for subsequent recommendation and search optimization. The search engine module includes basic search functions, commodity search, keyword matching and advanced functions, fuzzy search and natural language understanding to help users quickly find target commodities. The recommendation system module provides personalized commodity recommendations to users based on collaborative filtering, content recommendation and hybrid recommendation technologies to improve user experience and makes accurate recommendations in combination with user historical behavior, popular commodities and seasonal information. The intelligent assistant module provides a virtual secretary function to support users to interact with natural languages, query commodities, set reminders and manage shopping lists through voice or text input, and improve interaction efficiency. The data analysis module monitors and analyzes the user's behavior on the platform in real time, extracts user needs and trends, and optimizes search and recommendation strategies. The multimodal interaction module supports intelligent interaction in multiple input modes such as voice, text and picture, searches for commodities through voice, or uploads pictures for picture search, which brings convenient, fast and personalized service experience to consumers and also provides intelligent and efficient operation management for merchants. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings;

[0030] Figure 1 This is a schematic diagram of the overall framework of an e-commerce platform search recommendation and intelligent secretary integrated system of the present invention;

[0031] Figure 2 This is a structural schematic diagram of a computer device in an e-commerce platform search recommendation and intelligent secretary integrated system of the present invention. DETAILED DESCRIPTION

[0032] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] like Figure 1-2 As shown, an e-commerce platform search recommendation and intelligent secretary integrated system includes:

[0034] The user portrait module builds accurate user portraits by analyzing user behavior data and personalized preferences, providing a basis for subsequent recommendations and search optimization;

[0035] Search engine module, including basic search functions, product search, keyword matching and advanced functions, fuzzy search, natural language understanding, to help users quickly find target products;

[0036] The recommendation system module provides personalized product recommendations to users based on collaborative filtering, content recommendation, and hybrid recommendation technologies, improves user experience, and makes accurate recommendations based on user historical behavior, popular products, and seasonal information;

[0037] The intelligent assistant module provides a virtual secretary function, supports users to interact with products in natural language, and can query products, set reminders, and manage shopping lists through voice or text input, thus improving interaction efficiency.

[0038] Data analysis module, which monitors and analyzes user behavior on the platform in real time, extracts user needs and trends, and optimizes search and recommendation strategies;

[0039] The multimodal interaction module supports intelligent interaction with multiple input methods such as voice, text, and pictures. You can search for products by voice or upload pictures for image search.

[0040] Specifically, the user portrait module includes collecting user behavior data, browsing records, search keywords, click history, shopping cart items, purchase history, favorites, evaluations and feedback, obtaining basic information of users, collecting data from user login behaviors, social media associated accounts, and other third-party platforms, analyzing users' browsing, clicking, shopping frequencies and behavior patterns, judging users' interests and preferences for specific categories of goods, discovering users' active time, shopping cycles, and users' consumption habits during holidays and promotional periods through analysis of time periods, generating users' interest tags based on their historical data, dynamically adjusting interest weights based on users' click-through rates and dwell time on different types of goods, providing priority references for the recommendation system, clustering users with similar behavior patterns to form specific user groups, and grouping users to accept different types of recommendation strategies to improve the accuracy of recommendations, using natural language processing technology to analyze users' evaluations, feedback or customer service chat records, understand users' emotional tendencies, provide a basis for the platform to improve products and services, and push relevant after-sales services or problem solutions in a timely manner. As user behavior changes, the portrait will be updated in real time, and new purchase behaviors, evaluations, and browsing behaviors will be immediately incorporated into the portrait model to ensure that the system responds to user interests in a timely manner.

[0041] Specifically, the search engine module includes returning relevant product results based on the keywords input by the user, performing word segmentation processing on the natural language input by the user, establishing a search index, ensuring fast query and response, providing automatic correction or suggestions when the keywords input by the user are misspelled, calculating the relevance between the product and the search term based on the matching degree between the keyword and the product description, and sorting the search results based on the relevance, customizing the sorting of search results for each user in combination with user portraits, comprehensively considering multiple factors such as price, sales volume, evaluation, inventory, and brand, and sorting and optimizing the search results in multiple dimensions, understanding the user's complex queries through natural language processing technology, and for inaccurate or vague search terms, the system provides similar or related products. When the user enters a search term, the system automatically associates related words based on the existing search history and popular products to help the user quickly find the desired product, and quickly filter products by clicking on different categories based on the product classification information, and supports refined filtering by color, size, and material product attributes to help users quickly locate specific products, and recommend other products or product combinations related to the search in combination with the user's search keywords and behavior history.

[0042] Specifically, the search engine module also includes returning a list of products based on precise matching of keywords to ensure that core search needs are met, increasing the recall range through a variety of expansion methods, such as synonym expansion, semantic analysis, and user profiling, to ensure that users can find relevant products even in the case of inaccurate searches, and returning similar or identical products based on image recognition technology through user uploads of pictures, adjusting search results based on the user's geographic location, current activity, and time context information, and customizing the display of recommended products related to the current search term based on the user's historical search and purchase records, identifying the effectiveness of search results through user click, purchase, and dwell time behavior data, and continuously adjusting search sorting strategies through these feedbacks, optimizing user search experience through experiments on different search algorithms and sorting rules, ensuring that the presented products are more in line with user needs, and ensuring that the inventory, price, and promotional information of the products are synchronized to the search results in real time to prevent users from purchasing sold out or expired products.

[0043] Specifically, the recommendation system module includes collecting various behavioral data of users on the platform, including browsing records, search records, click histories, purchase records, and evaluations, cleaning, deduplicating, and normalizing the original data to ensure the quality and consistency of the data for subsequent model training and application of recommendation algorithms, based on user collaborative filtering, and according to the preferences of similar users, recommending the products that these users like to the current user; based on the similarity between the products, recommending similar products to the user; based on product attributes and user preferences, recommending products similar to the products that the user has browsed or purchased in the past; comprehensively using collaborative filtering, content recommendation, and rule recommendation methods, dynamically adjusting recommendation strategies according to different scenarios and user characteristics, using neural network models to process complex user behavior data, and mining deep user interest relationships from them to generate more accurate recommendations; closely integrating the recommendation system with the user portrait module, providing highly personalized recommendations based on the user's age, gender, geographic location, and interest tag information, and adjusting the recommended content in real time according to the user's current behavior and contextual information.

[0044] Specifically, the intelligent assistant module supports users to interact with the assistant through voice or text through natural language processing and understanding, understands the real needs of users through semantic analysis and intent recognition, supports multiple rounds of dialogue between users and assistants, and continuously maintains context information in complex problem situations. Based on the keywords entered by the user, the intelligent assistant can quickly help users search for related products and recommend products that meet the needs. Combined with user portraits, the intelligent assistant recommends products of interest to users during the interaction process. When a user asks about a product, the intelligent assistant provides detailed product descriptions, including specifications, user reviews, and price fluctuation information, provides users with current discount information and coupons, and predicts price changes of products to help users make purchasing decisions, help users quickly add products to shopping carts, view shopping cart contents, remind users to use coupons, and even guide users to complete settlement. The order status can be checked at any time through the intelligent assistant, including delivery progress and estimated delivery time. The intelligent assistant assists users in handling returns, exchanges, and complaints about after-sales issues, and guides users to complete operations according to the type of problem, providing users with automatic answers to common problems.

[0045] Specifically, the data analysis module includes collecting data from different sources, including user browsing behavior, search history, shopping history, evaluation data, logistics information, supporting data integration from multiple channels such as websites, apps, and mini-programs, and processing data in different formats, including user order records. User comments and customer service conversation data support the collection and processing of real-time data, including the user's current browsing, clicking, and searching behaviors, so as to facilitate timely adjustment of recommendation or promotion strategies, remove duplicate, abnormal or missing data, ensure the accuracy and consistency of data analysis, label and classify the collected data, and normalize the data, including time format unification and numerical standardization, to ensure that various types of data can be compared and analyzed with each other, analyze user interests and preferences based on user browsing, searching, and purchasing behaviors, form personalized user portraits, support accurate recommendations, divide users into different groups based on their behavioral characteristics, facilitate targeted marketing and promotion, identify users at risk of churn, predict which users will stop using the platform through behavioral patterns, and provide corresponding retention strategies, such as targeted discounts or reminder notifications, track the sales of different products, including sales volume, conversion rate, and inventory status, help merchants optimize product portfolios and replenishment strategies, and analyze user behavior. Evaluation and feedback of products, identify the advantages and disadvantages of products, help merchants improve product quality, optimize marketing strategies, track changes in product prices, analyze price fluctuations in different time periods or promotional periods, and competitors' pricing strategies, predict future sales trends based on historical sales data and market dynamics, help the platform to manage product inventory and predict the market, evaluate the effectiveness of various promotional activities, understand the impact of different promotional methods on sales, conversion rate, and user engagement, analyze the performance of different sales channels, optimize resource allocation, analyze the user's browsing path on the platform, understand the pages where users stay during the visit, bounce rate, and key nodes in the conversion funnel, help optimize the user experience, display the user's click distribution on the page through a heat map, identify key areas of user attention, help the platform optimize page design and content layout, analyze the entire process from user browsing to final ordering, identify links with low conversion rates, and make optimization suggestions.

[0046] Recommendation system effectiveness analysis: Evaluate the accuracy of the recommendation system and user satisfaction through user behavior data such as clicks and purchases on recommended products, and continuously optimize the recommendation algorithm.

[0047] Search popularity and trend analysis: Analyze the hot search terms, search times and trends on the platform to help identify users' immediate needs and market hotspots, and adjust product supply or promotion strategies in a timely manner.

[0048] Customized report generation: Generate customized sales, user, product, market and other data reports based on business needs, supporting statistical analysis by day, week, month or custom time period.

[0049] Data visualization tools: Provide visual chart tools, including bar charts, line charts, pie charts, heat maps, etc., to help operators understand data intuitively and make decisions quickly.

[0050] Real-time data dashboard: The dashboard displays key operational indicators (such as daily active users, real-time sales, etc.) to help management monitor business performance in real time.

[0051] Sales forecasting model: Through machine learning models, future sales are predicted based on historical data, helping e-commerce platforms to conduct scientific inventory management and market prediction.

[0052] User behavior prediction: By analyzing the user's historical behavior, we can predict future purchasing tendencies and the possibility of churn, helping the platform to formulate user retention or promotion plans in advance.

[0053] Product recommendation optimization: Through deep learning of user behavior data, we optimize the recommendation algorithm and improve product exposure and sales.

[0054] Data encryption and secure storage: Ensure that sensitive data such as user behavior data and personal information are encrypted and protected during collection, transmission and storage, in compliance with relevant privacy regulations.

[0055] Compliance review: Ensure that data analysis activities comply with privacy protection regulations such as GDPR, respect users' data privacy rights, and ensure that users have the right to delete or view their own data records. Through these functions, the data analysis module helps e-commerce platforms extract valuable business insights from massive data, optimize operational strategies, improve user experience, and drive sales growth.

[0056] Specifically, the data analysis includes analyzing the data in the database, including building a visualization analysis of data correlation and similarity functions using the inner product of the function, as follows:

[0057]

[0058] in,<f,g> is the inner product of function, f(x) is the similarity function of the data in the database, g(x) is the correlation function of the data in the database, a and b are real numbers representing the length of the data fragments intercepted in the database. The inner product of function is calculated on the data according to the length of the data in the database, the similar and related data are processed together, and the visual analysis is performed.

[0059] Specifically, the data analysis includes denoising the data in the database, including:

[0060]

[0061] Among them, λ is the universal threshold, σ is the database standard deviation, and N is the signal length.

[0062] Specifically, the data recording and analysis includes cleaning the data in the database, establishing a data matrix for the cleaned data, and classifying the cleaned data according to the data matrix, including

[0063]

[0064] Among them, X norm is the cleaned data, X is the original data size, and X max is the maximum value of the original data, X min The minimum value of the original data

[0065] Specifically, the multimodal interaction module includes multiple languages ​​input through voice or text, and provides corresponding responses to meet international needs. Through NLP technology, it understands the user's true intentions, supports multiple rounds of dialogue, searches by uploading pictures, identifies products based on image recognition technology and provides relevant recommendations, recognizes user gestures, expressions, and objects through the camera, and provides corresponding feedback based on the image content, recognizes user gesture commands through the camera, and users control the interface, slide products, and zoom in and out pictures through gestures. It supports interaction with the system through screen touch, is suitable for mobile devices and touch screen devices, and inputs questions through keyboard or touch Or command, the system understands user needs through text analysis technology and provides accurate replies. It presents replies in various forms, including text, pictures, and links, to provide a richer interactive experience. It combines information from multiple modalities, including voice, text, and images, to conduct comprehensive analysis and response, to ensure that the interaction process of different modalities is smooth and natural, and users can switch seamlessly between voice, image, and text modes. It analyzes users' emotions and feelings through voice, text, and images, adjusts the interaction method according to the user's emotional state, and provides personalized recommendations based on the user's different input modes. It supports users to query products or information through different input methods and conduct accurate searches in combination with other modes.

[0066] The system supports seamless switching between different devices. For example, users can search for products by voice on smart speakers and view product details on their mobile phones or computers.

[0067] Collaborative interactive experience: The user's input on one device can complement the operation on another device. For example, after searching for a product by taking a photo on a mobile phone, a user can view related video recommendations on a smart TV.

[0068] Shopping scenario: On e-commerce platforms, users can interact with the system through voice, images, touch, and other methods to quickly find the products they want.

[0069] Smart home scenarios: Users can control home devices through voice and gestures, such as adjusting the brightness of lights through voice or turning on and off the TV through gestures.

[0070] Education and training scenarios: In learning scenarios, users can ask questions through voice, and the system will answer them through voice, images or videos, providing an interactive learning experience.

[0071] Privacy data protection: When processing multimodal data, the system needs to ensure that the user's privacy data (such as voice recordings, images, etc.) is effectively protected, and all sensitive data is encrypted for storage and transmission.

[0072] User control: Users can view and delete data related to multimodal interaction at any time, such as voice records, image history, etc. Through these functions, the multimodal interaction module provides users with a more natural, intuitive and flexible way of interaction, enhances the intelligence and usability of the system, and is widely used in e-commerce, smart home, medical care, education and other fields, greatly improving user experience and platform efficiency.

[0073] Those skilled in the art will understand that Figure 2 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0074] In one embodiment, a computer-readable storage medium is provided, storing a computer program, which, when executed by a processor, implements the steps in the above-mentioned e-commerce platform search recommendation and intelligent secretary integrated system embodiment.

[0075] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program includes computer instructions, the computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the steps in the embodiment of the above-mentioned e-commerce platform search recommendation and intelligent secretary integrated system.

[0076] 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 related hardware through a computer program. 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, database or other media used in the embodiments provided in the present application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical storage, etc. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0077] The working principle of the integrated system of search recommendation and intelligent secretary of an e-commerce platform of the present invention is as follows: by analyzing the behavior data of users and their personalized preferences, an accurate user portrait is established; through basic search functions, product search, keyword matching and advanced functions, users are helped to quickly find target products; personalized product recommendations are provided to users based on collaborative filtering to improve user experience; a virtual secretary function is provided to support users to perform natural language interaction; real-time monitoring and analysis of user behavior on the platform, extraction of user needs and trends, and optimization of search and recommendation strategies; support for intelligent interaction of multiple input methods such as voice, text and pictures, product search by voice, or uploading pictures for image search, which brings convenient, fast and personalized service experience to consumers, and also provides intelligent and efficient operation management for merchants.

[0078] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. An e-commerce platform search recommendation and intelligent secretary integrated system, characterized in that: include: The user portrait module builds accurate user portraits by analyzing user behavior data and personalized preferences, providing a basis for subsequent recommendations and search optimization; Search engine module, including basic search functions, product search, keyword matching and advanced functions, fuzzy search, natural language understanding, to help users quickly find target products; The recommendation system module provides personalized product recommendations to users based on collaborative filtering, content recommendation, and hybrid recommendation technologies, improves user experience, and makes accurate recommendations based on user historical behavior, popular products, and seasonal information; The intelligent assistant module provides a virtual secretary function, supports users to interact with products in natural language, and can query products, set reminders, and manage shopping lists through voice or text input, thus improving interaction efficiency. Data analysis module, which monitors and analyzes user behavior on the platform in real time, extracts user needs and trends, and optimizes search and recommendation strategies; The multimodal interaction module supports intelligent interaction with multiple input methods such as voice, text, and pictures. You can search for products by voice or upload pictures for image search.

2. According to claim 1, an e-commerce platform search recommendation and intelligent secretary integrated system is characterized in that: The user portrait module includes collecting user behavior data, browsing history, search keywords, click history, shopping cart items, purchase history, favorites, evaluations and feedback, obtaining basic information of users, collecting data from user login behavior, social media associated accounts, and other third-party platforms, analyzing user browsing, clicking, shopping frequency and behavior patterns, judging user interest and preference for specific categories of goods, discovering user active time, shopping cycle and user consumption habits during holidays and promotional periods through analysis of time periods, generating user interest tags based on user historical data, dynamically adjusting interest weights based on user click-through rates and dwell time on different types of goods, providing priority references for the recommendation system, clustering users with similar behavior patterns to form specific user groups, and grouping users to accept different types of recommendation strategies to improve the accuracy of recommendations, using natural language processing technology to analyze user evaluations, feedback or customer service chat records, understand user emotional tendencies, provide a basis for the platform to improve products and services, and timely push relevant after-sales services or problem solutions. As user behavior changes, the portrait will be updated in real time, and new purchase behaviors, evaluations, and browsing behaviors will be immediately incorporated into the portrait model to ensure that the system responds to user interests in a timely manner.

3. According to claim 2, the e-commerce platform search recommendation and intelligent secretary integrated system is characterized in that: The search engine module includes returning relevant product results based on the keywords input by the user, performing word segmentation processing on the natural language input by the user, establishing a search index, ensuring fast query and response, providing automatic error correction or suggestions when the keywords input by the user are misspelled, calculating the relevance between the product and the search term based on the matching degree between the keyword and the product description, and sorting the search results based on the relevance, customizing the sorting of search results for each user in combination with user portraits, and comprehensively considering multiple factors such as price, sales volume, evaluation, inventory, and brand to sort and optimize the search results in multiple dimensions. Through natural language processing technology, the user's complex queries are understood. For inaccurate or vague search terms, the system provides similar or related products. When the user enters a search term, the system automatically associates related words based on the existing search history and popular products to help the user quickly find the desired product. According to the classification information of the product, the product can be quickly filtered by clicking on different categories, and refined filtering by color, size, and material product attributes is supported to help the user quickly locate specific products. In combination with the user's search keywords and behavior history, other products or product combinations related to the search are recommended.

4. According to claim 3, the e-commerce platform search recommendation and intelligent secretary integrated system is characterized in that: The search engine module also includes returning a list of products based on exact matching of keywords to ensure that core search needs are met. Through a variety of expansion methods, such as synonym expansion, semantic analysis, and user profiling, the recall range is increased to ensure that users can find relevant products even in the case of inaccurate searches. Through users uploading pictures, the search engine returns similar or identical products based on image recognition technology, adjusts search results based on the user's geographic location, current activity, and time context information, and customizes the display of recommended products related to the current search term based on the user's historical search and purchase records. The validity of search results is identified through user click, purchase, and dwell time behavior data, and the search sorting strategy is continuously adjusted through these feedbacks. Through experiments on different search algorithms and sorting rules, the user search experience is optimized to ensure that the presented products are more in line with user needs, and that the inventory, price, and promotion information of the products are synchronized to the search results in real time to prevent users from purchasing sold out or expired products.

5. According to claim 4, the e-commerce platform search recommendation and intelligent secretary integrated system is characterized in that: The recommendation system module includes collecting various behavioral data of users on the platform, including browsing records, search records, click histories, purchase records, and evaluations, cleaning, deduplication, and normalization of the original data to ensure data quality and consistency for subsequent model training and application of recommendation algorithms, based on user collaborative filtering, and according to the preferences of similar users, recommending products that these users like to the current user; recommending similar products to users based on the similarities between products; recommending products similar to products that users have browsed or purchased in the past based on product attributes and user preferences; comprehensively using collaborative filtering, content recommendation, and rule recommendation methods, dynamically adjusting recommendation strategies according to different scenarios and user characteristics, using neural network models to process complex user behavior data, and mining deep user interest relationships to generate more accurate recommendations; closely integrating the recommendation system with the user portrait module, providing highly personalized recommendations based on the user's age, gender, geographic location, and interest tag information, and adjusting recommended content in real time based on the user's current behavior and contextual information.

6. The e-commerce platform search recommendation and intelligent secretary integrated system according to claim 5, characterized in that: The intelligent assistant module supports users to interact with the assistant through voice or text through natural language processing and understanding, understands users' real needs through semantic analysis and intent recognition, supports users and assistants to conduct multiple rounds of dialogue, and continuously maintains context information in complex problem situations. Based on the keywords input by the user, the intelligent assistant can quickly help users search for related products and recommend products that meet the needs. Combined with user portraits, the intelligent assistant recommends products of interest to users during the interaction process. When users ask about a product, the intelligent assistant provides detailed product descriptions, including specifications, user reviews, and price fluctuation information, provides users with current discount information and coupons, and predicts price changes of products to help users make purchasing decisions, help users quickly add products to shopping carts, view shopping cart contents, remind users to use coupons, and even guide users to complete settlement. The order status can be checked at any time through the intelligent assistant, including delivery progress and estimated delivery time. The intelligent assistant assists users in handling returns, exchanges, and complaints about after-sales issues, and guides users to complete operations according to the type of problem, providing users with automatic answers to common problems.

7. The e-commerce platform search recommendation and intelligent secretary integrated system according to claim 6, characterized in that: The data analysis module collects data from different sources, including user browsing behavior, search history, shopping records, evaluation data, and logistics information. It supports data integration from multiple channels such as websites, apps, and mini-programs, and processes data in different formats, including user order records. User comments and customer service conversation data support the collection and processing of real-time data, including users' current browsing, clicking, and searching behaviors, so as to adjust recommendation or promotion strategies in a timely manner, remove duplicate, abnormal, or missing data, ensure the accuracy and consistency of data analysis, label and classify the collected data, and normalize the data, including unifying the time format and standardizing the values, to ensure that various types of data can be compared and analyzed with each other, analyze users' interests and preferences based on their browsing, searching, and purchasing behaviors, form personalized user portraits, support accurate recommendations, divide users into different groups based on their behavioral characteristics, facilitate targeted marketing and promotion, identify users at risk of churn, predict which users will stop using the platform through behavioral patterns, and provide corresponding retention strategies, such as targeted discounts or reminder notifications, track the sales of different products, including sales volume, conversion rate, and inventory status, help merchants optimize product portfolios and replenishment strategies, and analyze users' Evaluation and feedback of products, identify the advantages and disadvantages of products, help merchants improve product quality, optimize marketing strategies, track changes in product prices, analyze price fluctuations in different time periods or promotional periods, and competitors' pricing strategies, predict future sales trends based on historical sales data and market dynamics, help the platform to manage product inventory and predict the market, evaluate the effectiveness of various promotional activities, understand the impact of different promotional methods on sales, conversion rate, and user engagement, analyze the performance of different sales channels, optimize resource allocation, analyze the user's browsing path on the platform, understand the pages where users stay during the visit, bounce rate, and key nodes in the conversion funnel, help optimize the user experience, display the user's click distribution on the page through a heat map, identify key areas of user attention, help the platform optimize page design and content layout, analyze the entire process from user browsing to final ordering, identify links with low conversion rates, and make optimization suggestions.

8. The e-commerce platform search recommendation and intelligent secretary integrated system according to claim 6, characterized in that: The data analysis includes analyzing the data in the database, including building a visualization analysis of data correlation and similarity functions using the inner product of the function, as follows: in,<f,g> is the inner product of function, f(x) is the similarity function of the data in the database, g(x) is the correlation function of the data in the database, a and b are real numbers representing the length of the data fragments intercepted in the database. The inner product of function is calculated on the data according to the length of the data in the database, the similar and related data are processed together, and the visual analysis is performed.

9. The e-commerce platform search recommendation and intelligent secretary integrated system according to claim 8, characterized in that: The data analysis includes denoising the data in the database, including: Among them, λ is the universal threshold, σ is the database standard deviation, and N is the signal length.

10. According to claim 8, an e-commerce platform search recommendation and intelligent secretary integrated system is characterized in that the multimodal interaction module includes multiple languages ​​input through voice or text, and provides corresponding responses, adapts to international needs, understands the user's true intentions through NLP technology, supports multiple rounds of dialogue, searches by uploading pictures, identifies products based on image recognition technology and provides relevant recommendations, recognizes user gestures, expressions, and objects through the camera, and provides corresponding feedback based on the image content, recognizes user gesture commands through the camera, and users control the interface, slide products, and zoom in and out pictures through gestures. It supports interaction with the system through screen touch, and is suitable for mobile devices and touch Control the screen device and input questions or commands through the keyboard or touch. The system understands user needs through text analysis technology and provides accurate replies. The replies are presented in various forms, including text, pictures, and links, to provide a richer interactive experience. It combines multiple modal information, including voice, text, and images, to conduct comprehensive analysis and response, to ensure that the interaction process of different modalities is smooth and natural, and users can switch seamlessly between voice, image, and text modes. The system analyzes user emotions and feelings through voice, text, and images, adjusts the interaction method according to the user's emotional state, and provides personalized recommendations based on the user's different input modes. It supports users to query products or information through different input methods and conduct accurate searches in combination with other modes.

Images