Automatic commodity information complementing method based on semantic association

Abstract

The invention relates to the technical field of information analysis, and discloses a semantic association-based commodity information automatic completion method, which comprises the steps of obtaining all commodity information of a target commodity, judging whether the commodity information of the target commodity is missing or not, if the commodity information is missing, completing the commodity information, and if the commodity information is missing, completing the commodity information. According to the commodity information automatic completion method based on semantic association, through comprehensive analysis of the user behavior data and the purchase conversion data, whether the commodity information is missing or not is scientifically judged, subjective assume is avoided, effectiveness of completion measures is ensured, the specific part of information missing is accurately positioned, and the accuracy of commodity information completion is improved. A clear direction is provided for subsequent completion measures, blind completion is avoided, the requirements of the user for the commodity information are met by supplementing and perfecting the commodity information, the confusion and hesitation of the user when the user browses the commodity page are reduced, the user satisfaction is improved, and the purchase conversion rate is improved by optimizing commodity information display.

Description

Technical Field

[0001] The present invention relates to the technical field of information analysis, and in particular to a method for automatically completing product information based on semantic association. Background Art

[0002] With the booming development of e-commerce, product information, as a key factor in consumer decision-making, has become extremely important in terms of completeness and accuracy. Traditional methods of entering and maintaining product information often rely on manual work, which is inefficient, prone to errors, inconsistent information, and high maintenance costs. To address these issues, researchers and companies have begun to explore product information automatic completion technology. This technology uses technologies such as artificial intelligence and natural language processing to automatically obtain information from various sources and complete missing product information, thereby improving information quality and efficiency. The product information automatic completion method based on semantic association integrates advanced natural language processing technology, knowledge graph technology, and semantic association analysis algorithms to achieve smarter, more accurate, and more explainable automatic completion of product information, thereby improving the quality of product information, enhancing user experience, and providing more powerful data support for e-commerce platforms. The existing automatic completion method of product information based on semantic association cannot comprehensively analyze user behavior data and purchase conversion data to scientifically determine whether there is any missing product information. It is prone to subjective assumptions, resulting in invalid information completion and unable to accurately locate the specific part of the missing information, resulting in an inability to provide a clear direction for subsequent completion measures. It is prone to blind completion and cannot meet users' needs for product information. It increases users' confusion and hesitation when browsing product pages, reduces user satisfaction, and thus reduces purchase conversion rates. Its practicality has certain limitations. Summary of the Invention

[0003] The present invention provides a method for automatically completing product information based on semantic association, which is used to promote the solution of the problems raised in the background technology.

[0004] The present invention provides the following technical solution: a method for automatically completing product information based on semantic association, comprising: Get all product information of the target product and generate a partial set of product information: ; Determine whether the product information of the target product is missing; If the product information of the target product is not missing, the information completion operation will not be performed; If the product information of the target product is missing, the product information of the target product will be supplemented and the product information display strategy will be dynamically adjusted; The dynamic adjustment of product information display strategy is specifically as follows: For each part of the product information of the target product , calculate its interest weight: ; For each part of the product information of the target product , calculate its perplexity weight: ; For each part of the product information of the target product , according to its interest weight and confusion weight, determine the display content: .

[0005] As an optional solution of the method for automatically completing product information based on semantic association according to the present invention, determining whether the product information of the target product is missing includes sequentially analyzing all users who browsed the target product, specifically: Get Database ; Get the target product's ID, set it as the target ID, and record it as ; Extract all records of the target identifier in the database, denoted as ; Define a record analysis function to determine whether there is a browsing record for the target product: ; like , it is determined that there is a browsing record for the target product; Then query the user behavior database corresponding to the target identifier in the database, obtain all records corresponding to the target product in the user behavior database, and form a browsing record set: ; Generate a browsing user collection: ; like , it is determined that there is no browsing record for the target product; Then query the user behavior log corresponding to the target identifier in the database and generate a query result set: ; Set up a query analysis function to determine whether a browsing user set can be generated: ;like , then it is determined that a browsing user set can be generated; if , it is determined that the browsing user set cannot be generated; Generate a browsing user collection: .

[0006] As an optional solution of the method for automatically completing product information based on semantic association according to the present invention, wherein: all users browsing the target product are analyzed in sequence, the method further includes: Get browsing user collection ; For browsing user collections Each user in , extracting their detailed browsing behavior data when browsing the target product page; Define an attention function , calculate users For each product information section The degree of attention paid by users to each part of the product information during browsing is identified: ; Define a perplexity function , calculate users For each product information section The confusion level of each user is determined by the user, and the possible confusion points about each part of the product information are identified during the browsing process: ; Set up an attention analysis function to determine the user Focus when browsing the target product page: ; like , then determine the user Product information section Have a high level of attention; like , then determine the user Product information section There is a lower level of attention; Set up a perplexity analysis function to determine the user Possible confusion points when browsing the target product page: ; like , then determine the user In the product information section There may be confusion; like , then determine the user In the product information section There is no confusion; According to each product information section The result of the attention analysis function is for users Generate a set of focus points, denoted as : ; According to each product information section The result of the perplexity analysis function is for users Generate a set of confusion points, denoted as : ; Integrate a collection of concerns and confusion point set , generate user Browsing behavior analysis results.

[0007] As an optional solution to the semantic association-based product information automatic completion method of the present invention, determining whether the product information of the target product is missing includes comprehensively analyzing user browsing behavior data and purchase conversion process data to determine whether the product information of the target product is missing, specifically: Get browsing user collection ; Extract browsing user collection The users who add the target product to the shopping cart generate the set of users who add the product to the shopping cart, which is recorded as ; Calculate the add-to-cart rate of the target product, recorded as : ; Extract the users who enter the checkout page from the purchase-added user set and generate a checkout user set, recorded as ; Calculate the settlement rate of the target product, recorded as : ; Extract users who have successfully completed payment from the settlement user set and generate a payment user set, recorded as ; Calculate the payment rate of the target product, recorded as : ; Calculate the purchase conversion rate of the target product, recorded as : ; Statistics browsing user collection There is a set of confusion points in The number of users, calculate the proportion of confused users : ; Count the frequency of users' confusion points in key information parts : ; Get the best-selling product collection of the same type, recorded as : ; Calculate the difference between the target product and similar best-selling products in each part of the product information : ; Define a comprehensive analysis function to comprehensively determine whether the product information of the target product is missing: ; like , it is determined that the product information of the target product is missing; like , it is determined that the product information of the target product is not missing.

[0008] As an optional solution of the method for automatically completing product information based on semantic association according to the present invention, the product information of the target product is completed, including analyzing the user profile of the target product, specifically: Get browsing user collection ; Extract browsing user collection Each user in User portrait data: ; Based on browsing user collection Each user in User portrait data is used to group users: ; For each user group, calculate their attention to each part of the product information and analyze their focus and needs for the product information: ; According to the attention of each user group, the supplemented product information is generated for each part of the product information, which is recorded as : .

[0009] As an optional solution of the method for automatically completing product information based on semantic association described in the present invention, completing the product information of the target product also includes building and analyzing a user behavior model, specifically: Get browsing user collection ; Extract browsing user collection Each user in Behavioral data when browsing target product pages; For each user , defining its behavioral characteristic vector : ; For each user , constructing a behavioral path matrix : ; For each product information section , calculate its access frequency : ; Find out the key paths that appear most frequently in users' browsing behavior for target products : ; For each product information section , calculate its information demand , comprehensively considering the node access frequency and the behavioral characteristics on the critical path: ; Define a gap recognition function to determine the product information part Are there any information gaps? ; like , then determine the product information part There are information gaps; like , then determine the product information part There are no information gaps; According to the analysis results of each part of the product information under the gap identification function, the supplemented product information is generated for each part of the product information, which is recorded as : .

[0010] As an optional solution of the method for automatically completing product information based on semantic association described in the present invention, the completion of product information of the target product also includes mining user behavior association rules, specifically: Get browsing user collection ; Extract browsing user collection Each user in Behavioral data when browsing target product pages; For each user , define its behavior sequence: ; Each user The browsing behavior is regarded as a transaction, and a behavior transaction database is constructed. : ; Mining behavioral transaction databases Frequent itemsets in : ; The frequent itemsets are generated in the form of Association rules of Defining the confidence level of association rules : ; For association rules , define a gap check function to check whether there is missing information related to frequent subsequent behaviors in the product information: ; like , it is determined that there is an information gap in this part of the product information; like , it is determined that there is no information gap in this part of the product information; According to the association rules And the analysis results of each part of the product information under the gap check function, generate the supplemented product information for each part of the product information, recorded as : .

[0011] As an optional solution of the method for automatically completing product information based on semantic association described in the present invention, the completion of product information of the target product also includes cross-platform data integration and analysis, specifically: Set up a multi-channel data source collection: ; Integrate multi-channel data source collections and all user behavior data and feedback information to generate an integrated data set: ; Get browsing user collection ; Extract browsing user collection Each user in Behavioral data on browsing target products on different platforms; For each user , defining its behavior vectors on different platforms: ; Identify browsing user collections Hot topics among users; Extract hot topic collection ; For each user comment , analyze their emotional tendencies : ; Define a missing analysis function to determine the missing parts in the product information of the target product: ; like , then determine the product information of the target product Some parts are missing; like , then determine the product information of the target product Some parts are not missing; According to the analysis results of each part of the product information under the missing analysis function, the supplemented product information is generated for each part of the product information, which is recorded as : .

[0012] The present invention has the following beneficial effects: 1. This semantic association-based product information automatic completion method obtains all users who have browsed the target product by querying the user behavior database or user behavior log, and analyzes each user who has browsed the target product in turn through the refined browsing behavior data. This method covers all users who have browsed the target product, avoids missing any potential user feedback, accurately identifies the user's focus and confusion points during the browsing process, and analyzes each user's behavior to discover the differentiated needs of different user groups.

[0013] 2. This semantic association-based automatic product information completion method collects data from all aspects of the target product's purchase conversion process, including the add-to-cart rate, checkout page entry rate, payment success rate, purchase conversion rate, etc., determines the full-process purchase conversion data of the target product, comprehensively analyzes user browsing behavior data and the full-process purchase conversion data, determines whether there are any missing product information of the target product, comprehensively evaluates user behavior during the purchase process, discovers potential conversion bottlenecks, ensures that the completion measures can effectively improve the purchase conversion rate, scientifically determines whether there are any missing product information, avoids subjective conjecture, accurately locates the specific part of the missing information, and provides a clear direction for subsequent completion measures.

[0014] 3. This method of automatic product information completion based on semantic association collects user portrait data of target products, including basic information such as user age, gender, region, consumption level, purchase preferences, browsing history, evaluation history, etc. According to the user portrait data, it analyzes the differences in the concerns and needs of different user groups for product information, supplements and improves product information based on the concerns and needs of different user groups, meets users' needs for product information, improves users' experience when browsing product pages, reduces users' confusion and hesitation during the purchase process, thereby improving purchase conversion rates, continuously optimizing product information display, and ensuring that the information display effect is always in the best state.

[0015] 4. This method for automatically completing product information based on semantic association builds a user behavior model based on the browsing behavior data of a set of browsing users, describes the behavioral characteristics and paths of users when browsing the target product page, analyzes the key nodes and behavior patterns in the user behavior model, finds the information needs and possible information gaps of users during the browsing process, and supplements the missing parts of the product information based on the analysis results of the user behavior model to meet the user's demand for product information, improve the user's experience when browsing product pages, reduce the user's confusion and hesitation during the purchase process, thereby improving the purchase conversion rate, continuously optimizing the product information display, and ensuring that the information display effect is always in the best state.

[0016] 5. This method of automatic product information completion based on semantic association uses data mining technology, such as association rule mining algorithm, to mine the correlation between user behavior data. According to the mined association rules, it supplements the missing information related to key behaviors in the product information, meets the user's demand for product information, improves the user experience when browsing product pages, reduces the user's confusion and hesitation during the purchase process, thereby improving the purchase conversion rate, continuously optimizing the product information display, and ensuring that the information display effect is always in the best state.

[0017] 6. This semantic association-based automatic product information completion method integrates user behavior data and feedback information from multiple channels such as e-commerce platforms, social media platforms, and offline physical stores, analyzes the hot topics, concerns, and questions of users about the target products in cross-platform data, and supplements the missing parts of the product information based on the results of cross-platform data integration and analysis to meet users' demand for product information, improve the user experience when browsing product pages, reduce users' confusion and hesitation during the purchase process, thereby increasing the purchase conversion rate, and continuously optimizing the product information display to ensure that the information display effect is always in the best state.

[0018] 7. This semantic association-based automatic product information completion method dynamically adjusts the display method and content of product information according to the results of user behavior analysis, continuously optimizes the product information display, and ensures that the information display effect is always in the best state. By supplementing and improving product information, it meets the user's demand for product information and improves the user experience when browsing product pages. By optimizing the product information display, it reduces the user's confusion and hesitation during the purchase process, thereby improving the purchase conversion rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a flow chart of the method for automatically completing product information based on semantic association of the present invention. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making creative efforts are within the scope of protection of the present invention.

[0021] Example 1: A method for automatically completing product information based on semantic association, see Figure 1 ,include: Obtain all product information of the target product and generate a partial set of product information. The product information is information used to describe the specific situation of the product, including the product's picture, material, structure, etc.: ;in, Indicates the product information parts, ; Determine whether the product information of the target product is missing; If the product information of the target product is not missing, the information completion operation will not be performed; If the product information of the target product is missing, the product information of the target product will be supplemented and the product information display strategy will be dynamically adjusted; The dynamic adjustment of product information display strategy is specifically as follows: For each part of the product information of the target product , calculate its interest weight: ; in, Represents a user In the product information section The residence time, Represents a user In the product information section Number of clicks, Represents a user In the product information section The scroll depth, 、 、 are the weight coefficients of dwell time, click times, and scroll depth respectively, and , which is used to balance the impact of behavioral characteristics such as dwell time, number of clicks, and scroll depth on the interest weight. By adjusting these weights, the importance of certain behavioral characteristics can be emphasized, making them occupy a larger proportion in the calculation of interest weight. Represents a user In the product information section The residence time, Represents a user In the product information section Number of clicks, Represents a user In the product information section The scroll depth, 、 、 are the weight coefficients of dwell time, click times, and scroll depth respectively, and , which is used to balance the impact of behavioral characteristics such as dwell time, number of clicks, and scroll depth on the interest weight. By adjusting these weights, the importance of certain behavioral characteristics can be emphasized, making them occupy a larger proportion in the calculation of interest weight. Indicates the specific product information section that is currently being analyzed or considered for display strategy adjustment. It is the focus when calculating interest weight and is used to determine whether the display method of this section needs to be adjusted based on user behavior data. Represents all other parts of the product information, used as a comparison or normalization benchmark when calculating the interest weight. That is, it participates in the calculation as a reference system to ensure that the calculated interest weight is relative and comparable; For each part of the product information of the target product , calculate its perplexity weight: ; in, Represents a user In the product information section The number of back button clicks, Represents a user In the product information section Number of browsing path jumps, 、 are the weights of the number of back button clicks and the number of browsing path jumps, respectively, and , used to balance the impact of the number of back button clicks and the number of browsing path jumps on the perplexity. The contribution of different perplexity behaviors in the perplexity calculation can be adjusted according to actual needs. Represents a user In the product information section The number of back button clicks, Represents a user In the product information section Number of browsing path jumps, 、 are the weights of the number of back button clicks and the number of browsing path jumps, respectively, and , used to balance the impact of the number of back button clicks and the number of browsing path jumps on the perplexity. The contribution of different perplexity behaviors in the perplexity calculation can be adjusted according to actual needs; For each part of the product information of the target product , according to its interest weight and confusion weight, determine the display content: ; in, Indicates the product information part To highlight, Indicates the product information part To simplify the display, Indicates the product information part Display the original content. is the interest weight threshold, which is used to determine whether a product information section needs to be highlighted. When the interest weight of a product information section, that is, the relative value of the user's attention to this section relative to all product information sections, is greater than or equal to this threshold, the section is considered to be highly attractive to users and should be highlighted to further attract users' attention and enhance their attention to this section. The perplexity weight threshold is used to determine whether the product information section needs to be simplified or optimized. When the perplexity weight of the product information section, that is, the relative value of the user's confusion when browsing this section relative to all product information sections, is greater than or equal to this threshold, it is considered that there may be problems with this section, causing user confusion. In this case, consideration should be given to simplifying or optimizing the display content of this section to improve user experience and reduce user confusion when browsing this section.

[0022] Determining whether the product information of the target product is missing includes analyzing all users who browsed the target product in sequence, specifically: Get Database , the database A database that stores all products on the shopping platform, all product information for each product, all users of the platform, and each user's browsing and consumption information; Get the target product's ID, set it as the target ID, and record it as ; Extract all records of the target identifier in the database, denoted as ; Define a record analysis function to determine whether there is a browsing record for the target product: ; in, Indicates an empty set, that is, the target identifier has no record in the database, which means that the target product may not have a browsing record, or the browsing record of the target product is recorded in the user behavior log; like , it is determined that there is a browsing record for the target product; Then query the user behavior database corresponding to the target identifier in the database, obtain all records corresponding to the target product in the user behavior database, and form a browsing record set: ; in, is a query function used to identify the user behavior database corresponding to the target in the database. User behavior database corresponding to the target identifier; Generate a browsing user collection: ; in, Represents a user Target products Browsing behavior data, if the user Viewed target products , then the value is greater than or equal to 1, otherwise it is 0; like , it is determined that there is no browsing record for the target product; Then query the user behavior log corresponding to the target identifier in the database and generate a query result set: ; in, The user behavior log corresponding to the target identifier, is a query function used to query the user behavior log corresponding to the target identifier in the database; Set up a query analysis function to determine whether a browsing user set can be generated: ; like , then it is determined that a browsing user set can be generated; like , it is determined that the browsing user set cannot be generated; Generate a browsing user collection: ; in, Represents a user Target products Browsing behavior data, if the user Viewed target products , then the value is greater than or equal to 1, otherwise it is 0, Represents an empty set, that is, there is no browsing record for the target identifier and no record in the user behavior log, indicating that there is no browsing record for the target product.

[0023] The analysis of all users who browse the target product in sequence also includes: Get browsing user collection ; For browsing user collections Each user in , extracting detailed browsing behavior data when browsing the target product page, including but not limited to dwell time, browsing path, mouse movement trajectory, click location, number of clicks, scroll depth, number of images viewed, number of times details are expanded, etc.; Define an attention function , calculate users For each product information section The degree of attention paid by users to each part of the product information during browsing is identified: ; in, Represents a user In the product information section The residence time, Represents a user In the product information section Number of clicks, Represents a user In the product information section The scroll depth, which is the percentage of the page scrolled, 、 、 are the weights of dwell time, clicks, and scroll depth respectively, and , which is used to balance the impact of behavioral characteristics such as dwell time, number of clicks, and scroll depth on the interest weight. By adjusting these weights, the importance of certain behavioral characteristics can be emphasized. Represents browsing user collection The users, Indicates the product information parts; Define a perplexity function , calculate users For each product information section The confusion level of each user is determined by the user, and the possible confusion points about each part of the product information are identified during the browsing process: ; in, Represents a user In the product information section The number of back button clicks, Represents a user In the product information section Number of browsing path jumps, 、 are the weights of the number of back button clicks and the number of browsing path jumps, respectively, and , used to balance the impact of the number of back button clicks and the number of browsing path jumps on the perplexity. According to actual needs, the contribution of different perplexity behaviors in identifying perplexity points can be adjusted; Set up an attention analysis function to determine the user Focus when browsing the target product page: ; in, The attention threshold is used to determine whether the user's attention to a certain part of the product information is high enough, so as to decide whether the part should be highlighted or displayed first. like , then determine the user Product information section Have a high level of attention; like , then determine the user Product information section There is a lower level of attention; Set up a perplexity analysis function to determine the user Possible confusion points when browsing the target product page: ; in, The perplexity threshold is used to determine whether the product information section needs to be simplified or optimized. When the perplexity is greater than or equal to the perplexity threshold, it indicates that the user encountered great confusion when browsing the product information section. This may be because the information is unclear, incomplete, or too complicated, making it difficult for the user to understand or find the required content. In this case, this part of the content needs to be simplified or optimized to reduce user confusion and improve user experience. When the perplexity is less than the perplexity threshold, it indicates that the user's confusion level when browsing this part is within an acceptable range and no special adjustments are required. The original display method can be maintained. like , then determine the user In the product information section There may be confusion; like , then determine the user In the product information section There is no confusion; According to each product information section The result of the attention analysis function is for users Generate a set of focus points, denoted as : ; According to each product information section The result of the perplexity analysis function is for users Generate a set of confusion points, denoted as : ; Integrate a collection of concerns and confusion point set , generate user Browsing behavior analysis results.

[0024] In this embodiment, determining whether the product information of the target product is missing includes comprehensively analyzing the user browsing behavior data and the purchase conversion process data to determine whether the product information of the target product is missing, specifically: Get browsing user collection ; Extract browsing user collection The users who add the target product to the shopping cart generate the set of users who add the product to the shopping cart, which is recorded as ; Calculate the add-to-cart rate of the target product, recorded as : ;in, is the number of elements in the collection of users who added purchases. For browsing user collections The number of elements in Extract the users who enter the checkout page from the purchase-added user set and generate a checkout user set, recorded as ; Calculate the settlement rate of the target product, recorded as : ;in, The number of elements in the user collection for settlement; Extract users who have successfully completed payment from the settlement user set and generate a payment user set, recorded as ; Calculate the payment rate of the target product, recorded as : ;in, The number of elements in the collection of paying users; Calculate the purchase conversion rate of the target product, recorded as : ; Statistics browsing user collection There is a set of confusion points in The number of users, calculate the proportion of confused users : ;in, is the indicator function, if the user The set of puzzle points Not an empty set, that is, users There is a set of confusion points , then the value is 1, otherwise the value is 0. is the number of views; Count the frequency of users' confusion points in key information parts Key information refers to the parts of product information that have a significant impact on users' purchasing decisions and usage experience. These parts are usually the content that users pay most attention to when browsing product pages, and may also be the key factors that influence whether users purchase the product. They include product descriptions, product parameters, user reviews, applicable scenarios, after-sales service, price information, brand information, etc. ;in, The number of key parts of the target product information is the number of key information parts in the target product information, that is, the key information part set The number of elements in the product information reflects the number of parts that need to be focused on and optimized. express For the key information part, is the indicator function, if the user The set of puzzle points Contains key information sections , that is, the set of confusion points With key information section If there is overlap, the value is 1, otherwise the value is 0; Get the best-selling product collection of the same type, recorded as : ; Calculate the difference between the target product and similar best-selling products in each part of the product information : ; in, Indicates that the target product is in the product information Part of the content, Indicates similar best-selling products In the product information Part of the content, is an indicator function. If the target product is similar to the best-selling products In the product information If the content of the part is different, the value is 1, otherwise the value is 0. is the number of similar best-selling products. When calculating the difference between the target product and similar best-selling products in each part of the product information, we focus on the substantial difference in the content, rather than the literal exactness. We can judge whether the content is substantially different based on semantic similarity or information structure. Judging whether the content is substantially different based on semantic similarity is to use text similarity calculation methods (such as cosine similarity, Jaccard similarity, etc.) to measure whether the content of two products in a certain information part is similar. If the similarity is lower than a certain threshold, it is considered that the content is different. Judging whether the content is substantially different based on information structure is to compare whether the two product information parts contain the same fields or structured information. If there are missing fields or different structures, it is considered that the content is different. For example, suppose there is a target product and a similar best-selling product. , we need to compare them in the first The product information section includes function description, user reviews and applicable scenarios. The function description of the target product is "with high-speed processing capabilities and supports multi-tasking operation". Similar best-selling products The function description is "supports fast multi-tasking", and the two have high semantic similarity. It is determined that the two contents are the same, and the output is 1. The target product contains a review summary and detailed reviews in the user review content. Similar best-selling products The content of the user review only contains the review summary. The information structure of the two is different. If the content of the two is different, output 0. The content of the target product in the applicable scenario lists multiple usage scenarios. Similar best-selling products If the applicable scenario does not list the usage scenario in detail, and the sample being compared has missing content, the two samples are judged to have different content and the output is 0; Define a comprehensive analysis function to comprehensively determine whether the product information of the target product is missing: ; in, The confused user ratio threshold is used to measure whether the proportion of users who browse the target product page and are confused about the product information is too high. If the actual confused user ratio exceeds this threshold, it indicates that there may be problems with the product information and further analysis is needed to determine whether there is information missing. For example, if the confused user ratio threshold is 30%, The purchase conversion rate threshold is used to determine whether the ratio of the number of users who ultimately purchase the target product to the number of users who browse the product page is too low. A low purchase conversion rate may indicate that the product information is not attractive enough to users, or that key information is missing or unclear, affecting users' purchasing decisions. For example, if the purchase conversion rate threshold is 10%, The frequency threshold of confusion points is used to evaluate whether the frequency of confusion points when users browse key information parts is too high. When the frequency of confusion points exceeds this threshold, it means that the key parts of the product information may be unclear or missing, making it difficult for users to understand and requiring optimization. The difference threshold is used to compare the degree of difference in product information between the target product and similar best-selling products. If the difference exceeds the set threshold, it may mean that the target product has deficiencies in information display and needs to refer to the practices of best-selling products to supplement or optimize the information; like , it is determined that the product information of the target product is missing; like , it is determined that the product information of the target product is not missing; The key information collection process is as follows: For each product information section , calculate the browsing user set The ratio of the number of users who visited this section to the total number of users: ; For each product information section , calculate the browsing user set Average length of stay in this section: ; For each product information section , statistics browsing user collection Number of clicks and interactions in this section: ; in, For users In the product information section Number of clicks, For users In the product information section The number of interactive behaviors; Use TF-IDF algorithm to calculate each keyword TF-IDF value: ; in, ; ; Extract high-frequency keywords from user comments and discussions: ; in, TF-IDF screening threshold, used to extract high-frequency keywords from all keywords; Use topic modeling algorithms such as LDA to identify major topics in user comments and discussions: ; Analyze user behavior in the purchase conversion funnel and calculate the conversion rate at each stage: ; The purchase conversion funnel is a model used to describe the entire process from the user's initial contact with the product to the final completion of the purchase. It divides the user's purchasing behavior into multiple stages. Each stage has a corresponding conversion rate, which is used to measure the proportion of users who continue to advance to the next process in that stage. It usually includes browsing the product page, adding to the shopping cart, entering the checkout page, and completing the payment. The conversion rate refers to the ratio of the number of users who successfully complete a certain stage in the purchase conversion funnel to the total number of users who enter that stage. It is used to measure the user's conversion efficiency at that stage. The number of converted users is the number of users who successfully complete the actions in a certain stage. For example, in the "Complete Payment" stage, the number of converted users is the number of users who ultimately successfully pay. The total number of users is the total number of users who enter a certain stage. For example, in the "Browse Product Page" stage, the total number of users is the number of all users who have browsed the product page. Conduct A / B testing on different product information display methods to evaluate the impact of different information parts on conversion rate: ; Among them, A / B testing is a method that determines which design scheme is better by comparing the effects of two or more different design schemes. In product information display, A / B testing is used to evaluate the impact of different display methods on user behavior (such as conversion rate). Specifically, it includes: S1. Define the test goal, that is, clearly define the goal you want to achieve through A / B testing, for example, improving purchase conversion rate or increasing user stay time; S2. Select test variables, that is, determine the product information display method to be tested. For example, display method A is to place the "user evaluation" section of the product details page at the top of the page, and display method B is to place the "user evaluation" section of the product details page at the bottom of the page; S3. Create a test group and a control group, and randomly divide users into the test group and the control group to ensure that the two groups of users are statistically comparable. For example, the test group is display method A and the control group is display method B; S4. Implement the test, that is, display different product information display methods to the test group and the control group respectively within the same time period, and collect user behavior data, including browsing time, number of clicks, purchase conversion rate, etc.; S5. Calculate the conversion rate of the two groups of users respectively. The conversion rate calculation formula is , compare the conversion rates of the test group and the control group to evaluate the effectiveness of display method A and display method B; S6. Based on the test results, select the display method with better performance as the final solution; For each product information section , calculate its comprehensive score: ; in, 、 、 、 、 The weights assigned to each analysis result are based on business needs and meet , The specific formula is: ; in, and is a weight parameter used to balance the impact of keyword matching and topic similarity. The specific formula is: ; in, Is an indicator function, if the product information part If it contains high-frequency keywords, the value is 1, otherwise it is 0; The specific formula is: ; in, For product information The word frequency vector of For the theme The word frequency vector of is a vector and The dot product is used to measure the similarity of two vectors in direction. is a vector and The modulus length product is used for normalization to ensure that the similarity value is between -1 and 1; Based on the comprehensive score, determine the key information parts: ; in, is the weight threshold, which is used to determine the key information part. If the product information part If the comprehensive score of a part exceeds the weight threshold, it will be identified as the key information part.

[0025] Through the above method, by comprehensively analyzing user behavior data and purchase conversion data, we can scientifically judge whether there are any missing product information, avoid subjective conjectures, ensure the effectiveness of the completion measures, accurately locate the specific parts of the missing information, and provide a clear direction for subsequent completion measures to avoid blind completion. By supplementing and improving product information, we can meet users' demand for product information, reduce users' confusion and hesitation when browsing product pages, and improve user satisfaction. By optimizing the display of product information, we can reduce users' confusion and hesitation in the purchase process, thereby improving the purchase conversion rate. Through dynamic adjustment, we can continuously optimize the display of product information to ensure that the information display effect is always in the best state and adapt to market changes and dynamic changes in user needs.

[0026] Example 2: This example is an improvement made on Example 1. The method for automatically completing product information based on semantic association completes the product information of the target product, including analyzing the user profile of the target product. Specifically: Get browsing user collection ; Extract browsing user collection Each user in User profile data includes basic information such as age, gender, region, consumption level, as well as shopping-related information such as purchase preferences, browsing history, and evaluation history: ; in, Represents a user Age, Represents a user Gender, Represents a user area, Represents a user The consumption level, Represents a user purchasing preferences, Represents a user Browsing history, Represents a user 's evaluation history; Based on browsing user collection Each user in User portrait data is used to group users, that is, to divide users into different groups: ; in, is the number of clusters, represents the interval, Indicates the age range; For each user group, calculate their attention to each part of the product information and analyze their focus and needs for the product information: ;in, For user groups The number of users, Indicates the product information parts, The headquarters score of the product information. Represents a user Product information section The specific formula is: ;in, Represents a user In the product information section The residence time, Represents a user In the product information section Number of clicks, Represents a user In the product information section The scroll depth, which is the percentage of the page scrolled, 、 、 are the weights of dwell time, clicks, and scroll depth respectively, and , used to balance the impact of behavioral characteristics such as dwell time, number of clicks, and scroll depth on the interest weight. By adjusting these weights, the importance of certain behavioral characteristics can be emphasized; According to the attention of each user group, the supplemented product information is generated for each part of the product information, which is recorded as : ; in, The attention threshold is used to determine whether the user's attention to a certain part of product information is high enough. When the attention is greater than the threshold, it means that the user pays high attention to a certain part of product information. Otherwise, it means that the user pays low attention to a certain part of product information. Indicates detailed product information supplemented for user groups with high attention. Indicates the default content of product information.

[0027] This embodiment also provides for completing the product information of the target product and constructing and analyzing a user behavior model, specifically: Get browsing user collection ; Extract browsing user collection Each user in Behavioral data when browsing the target product page, including browsing time, number of clicks, scrolling depth, browsing path, etc. For each user , defining its behavioral characteristic vector : ; in, Represents a user No. Behavioral characteristics, such as browsing time, number of clicks, etc. For each user , constructing a behavioral path matrix : ;in, Represents a user From the product information section The probability of jumping to the next section, For product information parts, For users from Number of jumps, For users exist Total number of visits; For each product information section , calculate its access frequency : ;in, Represents a user Whether the product information section has been visited , where 1 means if the user Visited the product information section , then output 1, if the user The product information section has not been visited , then output 0; Find out the key paths that appear most frequently in users' browsing behavior for target products , critical path It is a set of frequently occurring behavioral paths, identified by the frequency of their occurrence: ;in, For the path, is the frequency of path occurrence, The threshold of the path frequency is used to filter the critical path set ,determine which paths are important enough in user browsing behavior to be considered critical paths; For each product information section , calculate its information demand , comprehensively considering the node access frequency and the behavioral characteristics on the critical path: ;in, Is the path importance, which means that in the critical path set The degree of influence of each path on user browsing behavior is as follows: ;in, Critical path set The path access frequency of a certain path in the data, that is, the number of times a certain path is visited by users, is the total number of paths, i.e. the set of browsing users The total number of browsing paths of all users in Critical path set The path length of a path in the , that is, the number of product information parts contained in a path, is the average path length, i.e., the browsing user set The average length of browsing paths of all users in ; Define a gap recognition function to determine the product information part Are there any information gaps? ;in, Indicates the product information section level of detail; if , then determine the product information part There are information gaps; like , then determine the product information part There are no information gaps; According to the analysis results of each part of the product information under the gap identification function, the supplemented product information is generated for each part of the product information, which is recorded as : ; in, To fill in the gaps in information, This is the default content of product information.

[0028] This embodiment also provides for completing the product information of the target product and mining user behavior association rules, specifically: Get browsing user collection ; Extract browsing user collection Each user in Behavioral data when browsing the target product page, including the sequence of product information sections visited; For each user , define its behavior sequence: ;in, Represents a user Visit the product information section, The number of product information sections visited for the user; Each user The browsing behavior is regarded as a transaction, and a behavior transaction database is constructed. : ; Each transaction Represents a partial set of product information accessed by a user, that is, each transaction Corresponding to a user's behavior sequence, it is expressed as ; Use association rule mining algorithms, such as the Apriori algorithm, to mine behavioral transaction databases Frequent itemsets in : ;in, The minimum support threshold is used to filter frequent item sets. The minimum support threshold sets a standard for distinguishing which item sets are frequent enough in user behavior data. Only item sets with a frequency exceeding the threshold will be retained and used as frequent item sets for further analysis. By setting this threshold, partial combinations of product information that are rarely accessed by users at the same time can be filtered out, thereby reducing the interference of noise data on the analysis, making subsequent association rule mining more focused on meaningful patterns, reducing the interference of unimportant item sets, and frequent item sets. The frequency of occurrence of all transactions in the behavioral transaction database is higher than the minimum support threshold Item set, is the support function, that is, the included item set The ratio of the number of transactions to the total number of transactions is as follows: ;in, Behavioral Transaction Database Contains The number of transactions, Behavioral Transaction Database Total number of transactions in ; The frequent itemsets are generated in the form of The association rules of and is a subset of the product information section, and ,Right now and are disjoint item sets; Defining the confidence level of association rules : ;in, is the minimum confidence threshold, which is used to evaluate the reliability of association rules. Therefore, association rules need to meet the minimum confidence threshold. ; For association rules , define a gap check function to check whether there is missing information in the product information that is related to frequent subsequent behavior. For example, if the association rule shows that users often check the applicable scenarios after viewing the product parameters, but the product information lacks a detailed description of the applicable scenarios, then it is considered that there is an information gap: ;in, Indicates product information The level of detail of the part, The detail threshold is used to determine whether the product information is detailed enough and whether more information needs to be added. like , it is determined that there is an information gap in this part of the product information; like , it is determined that there is no information gap in this part of the product information; According to the association rules And the analysis results of each part of the product information under the gap check function, generate the supplemented product information for each part of the product information, recorded as : ; in, Indicates the existence of a certain data or a certain condition. Indicates that a certain data or condition does not exist. To provide detailed information to address information gaps, This is the default content of product information.

[0029] This embodiment also provides for completing the product information of the target product and performing cross-platform data integration and analysis, specifically: Set up a multi-channel data source collection, which contains all platform channels related to the target product: ; Each data source Contains user behavior data and feedback information, including but not limited to browsing history, purchase history, likes, comments, shares, search keywords, etc.; Integrate multi-channel data source collections and all user behavior data and feedback information to generate an integrated data set: ; Get browsing user collection ; Extract browsing user collection Each user in Behavioral data on browsing target products on different platforms; For each user , defining its behavior vectors on different platforms: ;in, Represents a user In the Behavioral manifestations; Identify browsing user collections Hot topics among users; Extract hot topic collection ; For each user comment , analyze its emotional tendencies through sentiment dictionaries or machine learning models : ;in, Indicates user comments Positive comments, Indicates user comments A neutral comment. Indicates user comments Negative comments; Define a missing analysis function to determine the missing parts in the product information of the target product: ; in, For users to The specific formula for the partial demand is: ;in, and are the weights of access frequency and hot topic relevance, respectively, and , used to balance the importance of access frequency and hot topic relevance in demand calculation, The user's access frequency is: ;in, Represents a user Whether the product information section has been visited , output 1 if accessed, output 0 if not accessed, Represents browsing user collection The total number of users in the demand formula, The relevance of hot topics is as follows: ;in, is the number of comments related to the hot topic, that is, in the hot topic collection In the product information section The number of related comments, is the number of all comments, that is, the user comment set The total number of elements; in the demand formula, is the sentiment tendency weight, which is used to adjust the demand degree to more accurately reflect the user's actual demand for product information. The specific formula is: ;in, It's a comment The emotional tendency of 1 is positive, 0 is neutral, and -1 is negative. Is an indicator function, indicating the comment Whether it involves the product information part If you comment Product information section , then output 1, if the comment Does not involve product information , then the output is 0; in the missing analysis function, The level of detail used to measure the product information part The content completeness and richness of the content is as follows: ;in, Indicates the product information section The length of the text, such as the number of words or characters, Indicates the product information section The amount of multimedia content, such as the number of pictures, the number of videos, etc. and is a weight parameter used to balance the impact of text length and multimedia content quantity, satisfying ; like , then determine the product information of the target product Some parts are missing; like , then determine the product information of the target product Some parts are not missing; According to the analysis results of each part of the product information under the missing analysis function, the supplemented product information is generated for each part of the product information, which is recorded as : ; in, To provide detailed information to address information gaps, This is the default content of product information.

[0030] Among them, identify the browsing user collection Hot topics among users, specifically: For each data source , obtain all user reviews related to the target product in the data source to form a user review collection: ;in, Indicates the User comments, Indicates the total number of comments, each comment Contains review text, review time, user information, rating and other review related information; For each comment , extract keyword set : ;in, Indicates the Comments Keywords; For keywords , calculate its TF value: ; For keywords , calculate its IDF value: ; For keywords , calculate its TF-IDF value: ; According to the TF-IDF value, extract high-scoring keywords as a hot topic set : ; in, It is the threshold of TF-IDF value, which is used to distinguish which keywords have higher importance in the text collection. , you can filter out keywords whose TF-IDF values ​​are greater than or equal to the threshold. These keywords are generally considered to be words that contribute significantly to the text content.

[0031] This embodiment, through comprehensive analysis of user behavior data and purchase conversion data, scientifically determines whether there are any missing product information, avoids subjective assumptions, ensures the effectiveness of completion measures, accurately locates the specific parts of missing information, provides a clear direction for subsequent completion measures, and avoids blind completion. By supplementing and improving product information, it meets the user's demand for product information, reduces the confusion and hesitation of users when browsing product pages, and improves user satisfaction. By optimizing the display of product information, it reduces the confusion and hesitation of users in the purchase process, thereby improving the purchase conversion rate. Through dynamic adjustment, it continuously optimizes the display of product information to ensure that the information display effect is always in the best state and adapts to market changes and dynamic changes in user needs.

[0032] Example 3: This example is an example of all the above examples, that is, the overall algorithm of this application. Take a smartphone sold on Taobao as an example: Obtain smartphone product information based on the product ID through the Taobao product database or product information API, including product name, description, parameters (such as processor model, memory size, screen size, etc.), user reviews, applicable scenarios, after-sales service, etc. Query the Taobao user behavior database or log to obtain the unique identifiers of all users who have browsed the smartphone, form a browsing user set, and collect a list of users who have browsed the smartphone; For each user, detailed behavioral data is obtained when browsing the smartphone details page, such as dwell time, number of clicks on the parameter section and user review section, and scroll depth. Analysis and identification show that users pay more attention to the parameter and user review sections, but pay less attention to the applicable scenarios and after-sales service sections and are confused (such as frequently clicking the back button or jumping to other pages). The browsing behavior analysis results of each user show that many users spend a long time and click frequently on the parameter and user review sections, while showing confusion in the applicable scenarios and after-sales service sections. We collected data on the smartphone's purchase conversion process, including the add-to-cart rate, checkout page entry rate, payment success rate, and purchase conversion rate. We calculated that the add-to-cart rate was 50%, the checkout page entry rate was 80%, the payment success rate was 90%, and the purchase conversion rate was 36% (relatively low). Calculations revealed that 35% of users were confused (exceeding the 30% threshold). Analysis of the full purchase conversion process data revealed that the purchase conversion rate was lower than the expected 40% threshold. Combined with user browsing behavior data, we discovered that users were confused about the applicable scenarios and after-sales service. Comparing product information with similar best-selling smartphones revealed significant differences in the level of detail in the applicable scenarios and after-sales service sections for the target smartphone. This led us to conclude that the target smartphone had missing product information, primarily in the applicable scenarios and after-sales service sections. Comparing the product information of similar best-selling smartphones, we found that the target smartphones had significantly different levels of detail in terms of applicable scenarios and after-sales service compared to the best-selling products. Specifically: Suppose we selected three similar best-selling smartphones (S1, S2, and S3) for comparison. Their information in the applicable scenarios and after-sales service sections is as follows: the applicable scenarios of S1 include a detailed list of scenarios applicable to games, photography, office work, etc., and provide case descriptions; the after-sales service of S1 includes a 2-year warranty, and details the warranty scope, customer service response time, repair service process, etc.; the applicable scenarios of S2 cover a variety of usage scenarios, including games, travel, and daily office work, and emphasize its advantages; the after-sales service of S2 includes a 1-year warranty and a worry-free return and exchange policy, and details the return and exchange process and service outlets; the applicable scenarios of S3 include specific descriptions of scenarios applicable to games, video editing, multitasking, etc., with user scenario stories attached; the after-sales service of S3 includes a 3-year warranty, clear warranty terms, quick response service, and global warranty information; the target smartphone (denoted as T) has a simple applicable scenario. It is simply mentioned that it is applicable to multiple scenarios, but it is not expanded in detail. The target smartphone (denoted as T) only states that it provides a one-year warranty in the after-sales service section, lacking specific details. In the difference calculation, the difference in applicable scenarios is: compared with S1, T does not provide case descriptions and the content is brief; compared with S2, T does not emphasize the specific advantages of each scenario; compared with S3, T lacks user scenario stories. Summary of differences: T lacks details and specific cases in the description of applicable scenarios. The difference in after-sales service is: compared with S1, T has a shorter warranty period and does not clearly define the warranty scope and other details; compared with S2, T does not mention the worry-free return policy and service outlet information; compared with S3, T has a short warranty period and does not provide quick response service and global warranty information. Summary of differences: T lacks specific policies and service details in the after-sales service section. The difference between the target product and each similar best-selling product will be quantified, with a value of 1 indicating that there is a difference and 0 indicating that it is the same. The difference between the target product and each similar best-selling product is calculated: , where m=2, including two key information parts: applicable scenarios and after-sales services. For S1, , for S2, For S3, , then the final difference is ,Assuming that the threshold is set to 0.3, the difference is much higher than the threshold, which indicates that the ,level of details of the standard smartphone in the applicable scenarios and ,after-sales services is quite different from that of the best-selling ,products; User profile analysis: Analyzing the profiles of users who purchase this smartphone reveals that young users are more concerned with gaming performance and camera quality, while business professionals are more concerned with the compatibility of office applications and data security. Build a user behavior model: Based on user behavior data, we found that users often check user reviews after browsing the parameter section, but rarely browse the application scenarios and after-sales service sections and are confused. Mining user behavior association rules: Through association rule mining, it is found that users are more interested in applicable scenarios after viewing the parameters, but there is insufficient information about the current applicable scenarios. Cross-platform data integration and analysis: We integrated user discussions about mobile phones on social media platforms such as Weibo and found that users frequently discussed their experiences with mobile phones in different scenarios and after-sales service guarantees. Complete product information for the target product: For applicable scenarios, we added detailed descriptions and case studies of mobile phones in gaming, photography, office work, and other scenarios. For after-sales service, we detailed warranty policies, return and exchange procedures, customer service response times, and other information. This makes the completed smartphone product information more complete, better meeting the needs of different user groups and improving the user experience. Based on the feedback from the real-time data monitoring system and the results of user behavior analysis, it was found that users were more interested in the parameter part, so the display position of the parameter part was advanced and the key parameters were highlighted. For the applicable scenarios and after-sales service parts, after completing the information, the display strategy was dynamically adjusted, and the corresponding display area and guide signs were increased, so that the adjusted product information display is more in line with user needs, the user stay time is extended, and the purchase conversion rate is improved.

Claims

1. A method for automatically completing product information based on semantic association, characterized by: include: Get all product information of the target product and generate a partial set of product information: ； in, Indicates the product information parts; Determine whether the product information of the target product is missing; If the product information of the target product is not missing, the information completion operation will not be performed; If the product information of the target product is missing, the product information of the target product will be supplemented and the product information display strategy will be dynamically adjusted; The dynamic adjustment of product information display strategy is specifically as follows: For each part of the product information of the target product , calculate its interest weight: ； For each part of the product information of the target product , calculate its perplexity weight: ； in, Indicates the specific product information part that is currently being analyzed or considered for display strategy adjustment. It is the focus of attention when calculating the perplexity weight and is used to determine whether the display method of this part needs to be adjusted based on user behavior data. Represents all other parts of the product information, used as a benchmark for comparison or normalization when calculating the perplexity weight. That is, it participates in the calculation as a reference system to ensure that the calculated perplexity weight is relatively meaningful and comparable. For each part of the product information of the target product , according to its interest weight and confusion weight, determine the display content: 。 2. The method for automatically completing product information based on semantic association according to claim 1, characterized in that: Determine whether the target product's product information is missing, including analyzing all users who browsed the target product in sequence, specifically: Get Database ; Get the target product's ID, set it as the target ID, and record it as ; Extract all records of the target identifier in the database, denoted as ; Define a record analysis function to determine whether there is a browsing record for the target product: ； like , it is determined that there is a browsing record for the target product; Then query the user behavior database corresponding to the target identifier in the database, obtain all records corresponding to the target product in the user behavior database, and form a browsing record set: ； Generate a browsing user collection: ； like , it is determined that there is no browsing record for the target product; Then query the user behavior log corresponding to the target identifier in the database and generate a query result set: ； Set up a query analysis function to determine whether a browsing user set can be generated: ； like , then it is determined that a browsing user set can be generated; like , it is determined that the browsing user set cannot be generated; Generate a browsing user collection: 。 3. The method for automatically completing product information based on semantic association according to claim 2, characterized in that: Analyze all users who browse the target product in sequence, including: Get browsing user collection ; For browsing user collections Each user in , extracting their detailed browsing behavior data when browsing the target product page; Define an attention function , calculate users For each product information section The degree of attention paid by users to each part of the product information during browsing is identified: ； Define a perplexity function , calculate users For each product information section The confusion level of each user is determined by the user, and the possible confusion points about each part of the product information are identified during the browsing process: ； Set up an attention analysis function to determine the user Focus when browsing the target product page: ； Set up a perplexity analysis function to determine the user Possible confusion points when browsing the target product page: ; According to each product information section The result of the attention analysis function is for users Generate a set of focus points, denoted as : ; According to each product information section The result of the perplexity analysis function is for users Generate a set of confusion points, denoted as : ; Integrate a collection of concerns and confusion point set , generate user Browsing behavior analysis results.

4. The method for automatically completing product information based on semantic association according to claim 3, characterized in that: Determine whether the target product's product information is missing, including comprehensive analysis of user browsing behavior data and purchase conversion process data to determine whether the target product's product information is missing. Specifically: Get browsing user collection ; Extract browsing user collection The users who add the target product to the shopping cart generate the set of users who add the product to the shopping cart, which is recorded as ; Calculate the add-to-cart rate of the target product, recorded as : ; Extract the users who enter the checkout page from the purchase-added user set and generate a checkout user set, recorded as ; Calculate the settlement rate of the target product, recorded as : ; Extract users who have successfully completed payment from the settlement user set and generate a payment user set, recorded as ; Calculate the payment rate of the target product, recorded as : ; Calculate the purchase conversion rate of the target product, recorded as : ; Statistics browsing user collection There is a set of confusion points in The number of users, calculate the proportion of confused users : ; Count the frequency of users' confusion points in key information parts : ； Get the best-selling product collection of the same type, recorded as : ; Calculate the difference between the target product and similar best-selling products in each part of the product information : ； Define a comprehensive analysis function to comprehensively determine whether the product information of the target product is missing: ； like , it is determined that the product information of the target product is missing; like , it is determined that the product information of the target product is not missing.

5. The method for automatically completing product information based on semantic association according to claim 1, characterized in that: Complete the product information of the target product, including analyzing the user profile of the target product, specifically: Get browsing user collection ; Extract browsing user collection Each user in User portrait data: ； Based on browsing user collection Each user in User portrait data is used to group users: ； For each user group, calculate their attention to each part of the product information and analyze their focus and needs for the product information: ； According to the attention of each user group, the supplemented product information is generated for each part of the product information, which is recorded as : 。 6. The method for automatically completing product information based on semantic association according to claim 5, characterized in that: Completing the product information of the target product also includes building and analyzing user behavior models, specifically: Get browsing user collection ; Extract browsing user collection Each user in Behavioral data when browsing target product pages; For each user , defining its behavioral characteristic vector : ； For each user , constructing a behavioral path matrix : ； For each product information section , calculate its access frequency : ； Find out the key paths that appear most frequently in users' browsing behavior for target products : ； For each product information section , calculate its information demand , comprehensively considering the node access frequency and the behavioral characteristics on the critical path: ； in, Is the path importance, which means that in the critical path set The degree of influence of each path on user browsing behavior is as follows: ； Define a gap recognition function to determine the product information part Are there any information gaps? ； like , then determine the product information part There are information gaps; like , then determine the product information part There are no information gaps; According to the analysis results of each part of the product information under the gap identification function, the supplemented product information is generated for each part of the product information, which is recorded as : 。 7. The method for automatically completing product information based on semantic association according to claim 5, characterized in that: Completing the product information of the target product also includes mining user behavior association rules, specifically: Get browsing user collection ; Extract browsing user collection Each user in Behavioral data when browsing target product pages; For each user , define its behavior sequence: ； Each user The browsing behavior is regarded as a transaction, and a behavior transaction database is constructed. : ； Each transaction Represents a partial set of product information accessed by a user, expressed as ; Mining behavioral transaction databases Frequent itemsets in : ； The frequent itemsets are generated in the form of Association rules of Defining the confidence level of association rules : ； For association rules , define a gap check function to check whether there is missing information related to frequent subsequent behaviors in the product information: ； like , it is determined that there is an information gap in this part of the product information; like , it is determined that there is no information gap in this part of the product information; According to the association rules And the analysis results of each part of the product information under the gap check function, generate the supplemented product information for each part of the product information, recorded as : 。 8. The method for automatically completing product information based on semantic association according to claim 5, characterized in that: Completing the product information of the target product also includes cross-platform data integration and analysis, specifically: Set up a multi-channel data source collection: ； Integrate multi-channel data source collections and all user behavior data and feedback information to generate an integrated data set: ； Get browsing user collection ; Extract browsing user collection Each user in Behavioral data on browsing target products on different platforms; For each user , defining its behavior vectors on different platforms: ； Identify browsing user collections Hot topics among users; Extract hot topic collection ; For each user comment , analyze their emotional tendencies : ； Define a missing analysis function to determine the missing parts in the product information of the target product: ； like , then determine the product information of the target product Some parts are missing; like , then determine the product information of the target product Some parts are not missing; According to the analysis results of each part of the product information under the missing analysis function, the supplemented product information is generated for each part of the product information, which is recorded as : 。 9. The method for automatically completing product information based on semantic association according to claim 8, characterized in that: Identify browsing user collections Hot topics among users, specifically: For each data source , obtain all user reviews related to the target product in the data source to form a user review collection: ； For each comment , extract keyword set : ； For keywords , calculate its TF value: ； For keywords , calculate its IDF value: ； For keywords , calculate its TF-IDF value: ； According to the TF-IDF value, extract high-scoring keywords as a hot topic set : 。

Images