A personalized recognition and recommendation system and method based on a knowledge graph

Abstract

The present invention discloses a personalized recognition and recommendation system and method based on a knowledge graph, which relates to the field of personalized recommendation systems. It includes collecting historical behavior data of users and preprocessing it, constructing a cross-domain knowledge graph based on the preprocessed historical behavior data, generating a comprehensive user portrait, further analyzing the comprehensive user portrait to obtain a behavioral pattern analysis result, constructing a causal relationship model according to the comprehensive user portrait and the behavioral pattern analysis result, outputting the causal relationship between user behaviors, generating a personalized recommendation list for users based on the causal relationship between user behaviors, monitoring user behaviors in real time, and using an anomaly detection algorithm to analyze the difference between the latest behavior data of users and their historical behavior data. The present invention endows the user portrait with entity and relationship descriptions rich in semantics, significantly improves the personalization level of recommendations, can also identify potential interest points of users, and provides personalized recommendation results.

Description

Technical Field

[0001] The present invention relates to the field of personalized recommendation, and particularly to a personalized recognition recommendation system and method based on a knowledge graph. Background Art

[0002] With the rapid development of information technology and the Internet, personalized recommendation methods have gradually become an important means for various platforms to improve user experience. Traditional recommendation algorithms mainly rely on technologies such as collaborative filtering, content-based recommendation, and matrix factorization. These methods generate personalized recommendations by analyzing users' historical behaviors and the characteristics of products and content themselves.

[0003] However, there are still some deficiencies in the existing recommendation methods based on knowledge graphs during the construction and application processes. First, the construction of the knowledge graph is limited to a single domain, and the interactive behaviors of users in multiple domains cannot be fully captured, resulting in a one-sided user profile and being unable to comprehensively reflect the multi-dimensional interests of users. Second, existing knowledge graphs are often static and cannot be dynamically adjusted with the changes in user behaviors, resulting in the recommendation results lagging behind the real-time needs of users. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a personalized recognition recommendation system and method based on a knowledge graph to solve the problems that the construction of the knowledge graph is limited to a single domain, the interactive behaviors of users in multiple domains cannot be fully captured, resulting in a one-sided user profile and being unable to comprehensively reflect the multi-dimensional interests of users. Second, existing knowledge graphs are often static and cannot be dynamically adjusted with the changes in user behaviors, resulting in the recommendation results lagging behind the real-time needs of users.

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

[0007] In a first aspect, the present invention provides a personalized recognition recommendation method based on a knowledge graph, which includes collecting historical behavior data of users and preprocessing it;

[0008] Constructing a cross-domain knowledge graph based on the preprocessed historical behavior data, generating a comprehensive user profile, and further analyzing the comprehensive user profile to obtain a behavioral pattern analysis result;

[0009] Constructing a causal relationship model according to the comprehensive user profile and the behavioral pattern analysis result, and outputting the causal relationship between user behaviors;

[0010] Generating a personalized recommendation list for users based on the causal relationship between user behaviors;

[0011] Monitor user behavior in real time and use anomaly detection algorithms to analyze the differences between the user's latest behavior data and their historical behavior data;

[0012] Adjust the personalized recommendation list according to the difference results, and optimize the cross-domain knowledge graph and causal relationship model based on the user's behavior.

[0013] As a preferred solution of the personalized recognition and recommendation method based on the knowledge graph described in the present invention, wherein: collect the user's historical behavior data and preprocess it, specifically including the following steps,

[0014] Collect the user's click, browse, purchase, search, social interaction, and health data from multiple data sources;

[0015] Clean and deduplicate the collected data;

[0016] Process unstructured data using natural language processing techniques;

[0017] Standardize all data.

[0018] As a preferred solution of the personalized recognition and recommendation method based on the knowledge graph described in the present invention, wherein: construct a cross-domain knowledge graph based on the preprocessed historical behavior data, generate a comprehensive user portrait, and further analyze the comprehensive user portrait to obtain the behavior pattern analysis results, specifically including the following steps,

[0019] Use entity recognition technology to identify entities from the preprocessed data and identify the explicit and implicit relationships between entities;

[0020] Construct an independent knowledge graph for each domain, using the identified entities as nodes and the relationships between entities as edges;

[0021] Integrate the knowledge graphs of each domain into a unified cross-domain knowledge graph through entity alignment and relationship mapping technologies;

[0022] Based on the user nodes in the cross-domain knowledge graph, identify the relationships between the user nodes and other nodes and obtain the user's behavior data;

[0023] According to the user's behavior data, analyze the relationship strength between the user and the entity nodes in different domains, introduce recursive neural networks and multi-layer graph embedding technologies, recursively generate new graph levels after each user operation, capture the evolution of the user's behavior, and generate the user's dynamic comprehensive portrait;

[0024] Extract behavior characteristics in different domains based on the dynamic comprehensive portrait;

[0025] Analyze the user's behavior in different domains through the Apriori algorithm and find the associations between these behaviors;

[0026] Combine and analyze the behavioral characteristics and associations between behaviors in different fields of users to obtain behavioral analysis results.

[0027] As a preferred solution of the personalized recognition and recommendation method based on a knowledge graph according to the present invention, wherein: construct a causal relationship model based on the user comprehensive portrait and the behavioral pattern analysis results, and output the causal relationship between user behaviors, specifically including the following steps.

[0028] Extract key behavioral characteristics related to potential causal relationships from the user comprehensive portrait.

[0029] Take the key behavioral characteristics as the nodes of the Bayesian network.

[0030] Determine the causal relationship between the nodes through a data-driven learning method, and regard the causal relationship between the nodes as the edges of the Bayesian network.

[0031] Use maximum likelihood estimation to calculate the historical behavioral data to obtain the conditional probability table of each node and its parent node, and complete the construction of the Bayesian network.

[0032] Use the Bayesian network and generate the causal relationship before user behaviors based on the user comprehensive portrait and the behavioral pattern analysis results.

[0033] As a preferred solution of the personalized recognition and recommendation method based on a knowledge graph according to the present invention, wherein: generate a personalized recommendation list for the user based on the causal relationship between user behaviors, specifically including the following steps.

[0034] Extract the characteristics representing user behaviors from the user behavioral pattern analysis results to form the user's behavioral feature vector.

[0035] Calculate according to the causal relationship model and the user's behavioral feature vector to obtain the possibility of purchasing a commodity under the user's current behavioral pattern.

[0036] Based on the causal relationship between user behaviors, obtain the future needs of the user, and generate a personalized recommendation list with a priority ranking, and its expression is:

[0037]

[0038] wherein, Q i represents the priority of recommending commodity i, represents the possibility of purchasing commodity i under the user's current behavioral pattern, A u represents the behavioral feature vector of user u, V i represents the weight of commodity i, V j represents the weight of commodity j, C jDenote the actual interaction behavior of the user purchasing product j, C i Denote the actual interaction behavior of the user purchasing product i, and B represents the total number of products in the current product pool.

[0039] As a preferred solution of the personalized recognition and recommendation method based on a knowledge graph according to the present invention, wherein: the user behavior is monitored in real time, and an anomaly detection algorithm is used to analyze the difference between the user's latest behavior data and his historical behavior data, which specifically includes the following steps

[0040] Set a data difference threshold based on the historical behavior data;

[0041] Define a time window for real-time behavior;

[0042] Use the difference metric algorithm of time series to calculate the difference value between the real-time behavior data and the historical behavior data, and compare it with the difference threshold;

[0043] When the difference value is greater than the difference threshold, analyze these differences and mark the abnormal behavior, and further analyze the cause of the abnormal behavior.

[0044] As a preferred solution of the personalized recognition and recommendation method based on a knowledge graph according to the present invention, wherein: adjust the personalized recommendation list according to the difference result, and optimize the cross-domain knowledge graph and causal relationship model according to the user's behavior, which specifically includes the following steps

[0045] According to the difference detection result, identify the difference type between the user behavior and the historical behavior;

[0046] Adjust the personalized recommendation list based on the dynamically updated user behavior portrait and difference type;

[0047] Adjust and update the cross-domain knowledge graph and causal relationship model according to the real-time behavior data.

[0048] In a second aspect, the present invention provides a personalized recognition and recommendation system based on a knowledge graph, including

[0049] A data collection module that collects the historical behavior data of the user as the basic data for subsequent analysis;

[0050] A preprocessing module that cleans, denoises, formats, and normalizes the collected user behavior data;

[0051] A knowledge graph construction module that constructs a cross-domain knowledge graph based on the preprocessed historical behavior data to capture entity and their association relationships;

[0052] A user portrait module that combines the user's behavior data with the knowledge graph to generate a comprehensive portrait of the user;

[0053] The user behavior analysis module deeply analyzes the user's behavior patterns based on the user's comprehensive portrait, extracts the rules and preference patterns of the user's behavior;

[0054] The causal model construction module constructs a causal relationship model between user behaviors according to the user's comprehensive portrait and the analysis results of the behavior patterns, and identifies the causal chain behind the behaviors;

[0055] The personalized recommendation module generates a personalized recommendation list for the user based on the causal relationship model of the user's behavior, and matches its current needs and interests;

[0056] The monitoring and anomaly detection module monitors the user's latest behavior data in real time, and uses anomaly detection algorithms to analyze the differences between the latest data and the historical behavior data;

[0057] The list adjustment module dynamically adjusts the personalized recommendation list according to the results of real-time behavior and anomaly detection;

[0058] The knowledge graph and model optimization module continuously optimizes the domain knowledge graph and the causal relationship model according to the user's latest behavior data.

[0059] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the personalized recognition and recommendation method based on the knowledge graph as described in the first aspect of the present invention is implemented.

[0060] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the personalized recognition and recommendation method based on the knowledge graph as described in the first aspect of the present invention is implemented.

[0061] The beneficial effects of the present invention are as follows: By constructing a cross-domain knowledge graph based on data from multiple domains and generating a comprehensive portrait of the user, a comprehensive understanding of the user's multi-dimensional and multi-domain behaviors is achieved. Integrating and analyzing the behavior characteristics of the user in different domains endows the user portrait with semantically rich entity and relationship descriptions, significantly improving the personalization level of recommendations. By dynamically adjusting the recommendation list, the recommendation results can be updated in a timely manner when the user's behavior changes, thus ensuring the real-time and accuracy of the recommendations. By introducing the causal relationship model, not only can the current needs of the user be met, but also the potential interest points of the user can be identified, providing personalized recommendation results, greatly improving the user experience and increasing the rationality and accuracy of the recommendations. Description of the Drawings

[0062] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0063] Figure 1 It is a flowchart of the personalized recognition and recommendation method based on the knowledge graph in Embodiment 1.

[0064] Figure 2 It is a system diagram of the personalized recognition and recommendation system based on the knowledge graph in Embodiment 1. Detailed implementation manners

[0065] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific implementation manners of the present invention in conjunction with the accompanying drawings of the specification.

[0066] In the following description, many specific details are set forth to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar promotions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0067] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that mutually excludes other embodiments.

[0068] Embodiment 1, referring to Figure 1 and Figure 2 , is the first embodiment of the present invention. This embodiment provides a personalized recognition and recommendation method based on the knowledge graph, including the following steps:

[0069] S1. Through multi-channel (such as e-commerce platforms, social media, health applications, financial fields, etc.) and multi-dimensional data collection, various behaviors of users can be covered to ensure the comprehensiveness and diversity of data.

[0070] S1.1. The historical behavior data collected includes but is not limited to click data, browsing data, purchase data, search data, social interaction data, health data, etc.

[0071] Click data refers to the specific objects (such as products, articles, videos, etc.) clicked by users on the platform, reflecting the immediate interests of users.

[0072] Browsing data refers to the time users spend on a page and the content they view, which is used to analyze the depth of users' interests and preferences.

[0073] Purchase data refers to users' purchase history, including information such as the purchased goods, time, and amount, which directly reflects users' consumption habits and demands.

[0074] Search data refers to users' search behaviors on the platform, which reveals users' active demands and emerging interests.

[0075] Social interaction data refers to users' interaction behaviors on social platforms (such as likes, comments, shares, etc.), which is used to build users' social relationship networks, analyze users' social influence and its potential impact on behaviors.

[0076] Health data refers to, especially in health or medical platforms, collecting users' health data (such as physical examination records, exercise data, etc.), which helps to provide personalized health advice and recommendations.

[0077] Furthermore, by collecting users' behavior data from multiple data sources, it is possible to comprehensively and stereoscopically obtain users' behavior information, not limited to data in a specific dimension, which greatly enriches the dimensions of the user profile. This provides comprehensive data support for the subsequent cross-domain knowledge graph construction and user behavior analysis, enables a more accurate understanding of users' diverse needs and interests, and improves the personalization degree and accuracy of recommendations.

[0078] S1.2. Clean and deduplicate the collected data.

[0079] In users' historical behavior data, unstructured data (such as texts, comments, search terms, product descriptions, etc.) occupies a large proportion. To make full use of this data, natural language processing technology is introduced for analysis and processing.

[0080] The processing process includes text analysis, entity recognition, and sentiment analysis.

[0081] Text analysis refers to performing operations such as word segmentation, part-of-speech tagging, and sentiment analysis on text data such as users' comments and descriptions, and extracting useful information in the data, such as users' specific needs and sentiment tendencies.

[0082] Entity recognition refers to identifying key entities (such as products, brands, locations, etc.) in text through natural language processing technology, so as to convert unstructured data into structured data for subsequent knowledge graph construction.

[0083] Sentiment analysis refers to performing sentiment analysis on users' comments and feedback, and identifying users' attitudes towards a certain product or service, so as to better understand users' preferences.

[0084] Furthermore, through natural language processing technology, it is possible to effectively mine the deep - level information in unstructured data, complementing the deficiencies of structured data. This processing process improves the utilization rate of data, can more accurately identify features such as users' interest points and sentiment tendencies, and thus provides a richer basis for personalized recommendations. At the same time, natural language processing technology converts unstructured data into structured data, providing more semantic information for the construction of knowledge graphs, enhancing the detail richness of knowledge graphs and the accuracy of recommendations.

[0085] S1.4. To ensure that data from different data sources can be effectively compared and analyzed within the same framework, all data has been standardized.

[0086] S2. Extract key entities from the pre - processed historical behavior data through entity recognition technology. Entities can be users, products, content, locations, brands, etc., depending on the application scenario.

[0087] Entity recognition technology relies on natural language processing algorithms and automatically identifies important entities in the data by combining context information. This includes not only explicit entities (such as product names, user IDs, etc.) but also implicit entities (such as users' interests, sentiment tendencies, etc.). It can also identify explicit relationships (such as direct purchases, browsing, etc.) and implicit relationships (such as potential interest associations, emotional resonances, etc.) between these entities through context.

[0088] S2.1. After identifying entities and their relationships, independent knowledge graphs are constructed for each domain (such as e - commerce, social, health, etc.). Each graph consists of entities (nodes) and relationships (edges) between entities.

[0089] For example, in the e - commerce domain, entities such as products, brands, and users establish relationships through behaviors such as browsing, commenting, and purchasing; in the social domain, the interaction behaviors between users and content, and users and social friends form relationships. The purpose of independently constructing domain - specific knowledge graphs is to ensure that the semantic information of each domain can be fully mined and expressed, avoiding data confusion between different domains.

[0090] Furthermore, by constructing independent knowledge graphs for each domain, it is possible to capture in detail the behavior characteristics of users in different domains, avoiding misunderstandings and biases caused by simple data mixing. This fine - grained division can generate more accurate and context - relevant recommendations for each domain, effectively improving the accuracy of recommendation results.

[0091] S2.2. Integrate knowledge graphs from different fields into a unified cross-domain knowledge graph through entity alignment and relationship mapping techniques. Entity alignment technology identifies the same or similar entities (such as users, brands, etc.) in different fields and maps them to the same node. Relationship mapping technology identifies the commonalities of relationships in different fields (such as the connection between purchase and browsing behavior) and standardizes them into a unified relationship.

[0092] Further explanation: By integrating independent domain graphs into a unified cross-domain knowledge graph, a global perspective analysis of user behavior can be achieved. This can not only analyze user behavior in a single domain, but also understand the linkage effect of user behavior in different domains. Cross-domain graph integration greatly enhances semantic understanding capabilities, making recommendation results more context-aware and providing richer and more comprehensive recommendations.

[0093] S2.3. In a unified cross-domain knowledge graph, users exist as core nodes. By analyzing the relationship between user nodes and other entity nodes, user behavior data is extracted. These relationships can be direct (such as a user purchased a certain product) or indirect (such as a user's friend recommended a certain product). Not only does it focus on the relationship between users and a single entity, but it also analyzes the complex interactions between users and multiple entities.

[0094] It is further explained that by extracting user behavior data from cross-domain knowledge graphs, it is possible to accurately capture the user's interactions in multiple fields. This comprehensive and detailed behavior data provides important support for generating a comprehensive profile of the user, and can make recommendations based on the user's global behavior, further improving the relevance of recommendations and user satisfaction.

[0095] S2.4. Calculate the relationship strength between users and entities in different fields. The expression is:

[0096]

[0097] in, Represents user u and entity e i The strength of the relationship within time t, N represents the number of interaction types between the user and the entity (such as clicks, views, etc.), and w k represents the weight of the k-th type of interaction behavior, f k (u,e i ,t') represents the interaction between user u and entity e at time t' i The frequency of the k-th type of interaction behavior, α k Represents the attenuation coefficient of the kth type of behavior, which controls the decay rate of the interactive behavior over time. t represents the current time point, and t' represents the past time point.

[0098] S2.4.1. Based on knowledge graphs in different domains, where users interact with entity nodes in multiple domains, normalize these interaction relationships and calculate the comprehensive relationship metric between the user and the entity. The expression is as follows:

[0099]

[0100] Among them, R u (e i ) represents the comprehensive relationship metric between user u and entity e i . M represents the number of domains, and β j represents the weight of domain j, reflecting the importance of this domain in the overall relationship;

[0101] S2.4.2. Construct a feature function g(e i ) based on the multi-dimensional attribute features of entity e i to reflect the impact of different aspects of the entity on the user's comprehensive portrait;

[0102] Combine recursive neural networks and multi-layer graph embedding techniques to recursively generate the user's dynamic comprehensive portrait based on the data updated by each user behavior. The expression is as follows:

[0103]

[0104] Among them, represents the dynamic comprehensive portrait generated after the (n + 1)-th operation of user u. L represents the number of entities in the domains involved by the user, and γ i represents the behavior weight of the user in the domain where entity e i is located. represents the comprehensive relationship metric between the user and entity e i at the n-th operation. g(e i ) represents the impact of the entity's features on the generation of the user portrait. λ represents the time decay parameter, controlling the influence weight of recent behaviors on the user portrait. T(e i ) represents the latest interaction time between the user and entity e i . de i represents the variable for integrating entity e i , used to represent the continuous cumulative contribution to the entity.

[0105] The dynamic comprehensive portrait has a value range of [0, ∞). The larger the value, the stronger the user's interest, and the more likely they are to receive recommendations in these domains.

[0106] Furthermore, the dynamic comprehensive portrait generated by introducing recursive neural networks and multi-layer graph embedding technology can provide a comprehensive understanding of the user's interests, needs, consumption habits, etc., and can also capture and analyze changes in user behavior in real time to ensure the dynamics and timeliness of the recommendation results.

[0107] S2.5. Extract the user's behavioral characteristics in different fields based on the dynamic comprehensive portrait, including field preferences, behavior frequencies, and behavior patterns.

[0108] The field preference refers to the interest distribution of the user in multiple fields. For example, the user may have more purchase behaviors in the e-commerce field and more interaction behaviors in the social field.

[0109] The behavior frequency refers to the frequency of occurrence of the user's behavior in a specific field. For example, in the health field, the frequency of recording the user's step count and heart rate data every day.

[0110] The behavior pattern means that the user's behavior patterns may show similar trends in different fields. For example, the user's behavior of browsing and purchasing goods on the e-commerce platform may be related to their sharing behavior on social media.

[0111] S2.6. Define the user's behaviors in different fields as items based on the dynamic comprehensive portrait. For example, the user's browsing behavior in the e-commerce field, the like behavior in the social field, the step count exceeding 5000 steps in the health field, etc.

[0112] S2.6.1. Apply the Apriori algorithm to find frequently occurring behavior combinations. For example, the behavior of the user browsing goods in the e-commerce field and the behavior of the user liking in the social field may often occur simultaneously.

[0113] The Apriori algorithm is a classic association rule mining algorithm, often used to analyze frequent item sets and association rules between item sets in large datasets. Its core idea is to iteratively discover frequent item sets and then generate association rules based on these frequent item sets.

[0114] S2.6.2. Generate association rules between behaviors based on the frequent item sets. For example, the rule Indicates that after browsing the goods, the user usually likes the goods on the social platform.

[0115] Further explanation, through the analysis of the Apriori algorithm, potential associations between a user's behaviors in different fields can be identified. This association analysis not only helps to understand the user's behavior patterns but can also be used to optimize the recommendation function. For example, by identifying the association between a user's purchase behavior in the e-commerce field and sharing behavior in the social field, relevant products can be recommended for purchase when the user shares a certain type of product.

[0116] S2.7 Combine the analysis of the user's behavior characteristics and association rules to obtain behavior patterns such as cross-field behavior transfer, behavior prediction, and personalized recommendation optimization.

[0117] The cross-field behavior transfer refers to identifying the behavior transfer patterns of a user in different fields. For example, through association rules It is speculated that the user's behavior in the e-commerce field may affect their behavior in the social field.

[0118] The behavior prediction means that based on the user's current behavior characteristics and association rules, the user's future behavior can be predicted. For example, if a user shares a certain type of product on a social platform, it is speculated that the user may purchase similar products on an e-commerce platform.

[0119] The personalized recommendation optimization refers to optimizing the recommendation function by identifying the user's behavior patterns and association rules to provide more personalized recommendations for the user. For example, if a user often browses a certain type of product and likes similar products on a social platform, these products can be preferentially recommended.

[0120] Further explanation, through the Apriori algorithm, potential associations between a user's behaviors in different fields can be discovered. This cross-field behavior analysis can combine the user's behavior in one field to predict their potential needs in other fields. By combining the analysis of the user's behavior characteristics and association rules, the user's behavior patterns can be mined.

[0121] S3. Based on the comprehensive portrait, analyze the user's behaviors such as clicks, browsing, purchases, and searches to identify the key behavior characteristics that affect subsequent behaviors. For example, if a user frequently searches for a certain type of product recently, it may indicate a tendency to purchase soon.

[0122] Key behavior characteristics are those that have a significant association with the user's behavior decision-making or subsequent behavior changes.

[0123] For example, in the case of purchase behavior, a user's purchase of a certain type of product may be the result of other behaviors (such as browsing, clicking) or trigger subsequent behaviors (such as evaluation, repeat purchase).

[0124] In the case of click behavior, a user's click on a product on an e-commerce platform may be affected by previous search behaviors and may also affect subsequent purchase decisions.

[0125] Search behavior. A user's search behavior may reveal potential interests or needs and is often a precursor to purchasing behavior.

[0126] S3.1. Use the key behavior features extracted from the user's comprehensive profile as nodes in the Bayesian network.

[0127] Each node represents a user behavior feature, such as browsing product A, searching for product B, purchasing product C, etc. The directed edges between the nodes represent causal relationships, that is, the occurrence of one behavior (node) may affect subsequent behaviors. For example, a user's click on product A may lead to subsequent purchase of product A.

[0128] S3.2. After determining the nodes, gradually add or delete edges through greedy search, and evaluate the impact of each modification on the structure score by incremental calculation of BIC. In each candidate network structure, use BIC to calculate the score of the structure, and select the network structure with the highest score as the final Bayesian network structure.

[0129] The BIC refers to the Bayesian Information Criterion, which is a widely used scoring criterion. Its goal is to penalize overly complex models while ensuring that the model fits the data.

[0130] The final Bayesian network structure directly reflects the dependencies and potential causal relationships between user behavior features.

[0131] S3.2.1. Determine which behavior features (nodes) have causal relationships through the final Bayesian network structure, and determine the direction of the causal relationship (i.e., which behavior may cause the occurrence of another behavior).

[0132] For example, the final network structure may contain the following causal relationships:

[0133] User searches for product A -> User browses product A

[0134] This edge indicates that after a user searches for a certain product, they may browse the details page of that product, which is a reasonable causal relationship.

[0135] User browses product A -> User purchases product A

[0136] This edge indicates that after a user browses a certain product, they may further purchase that product, which is another reasonable causal relationship.

[0137] S3.2.2. After determining the structure of the network, calculate the conditional probability table for each node based on historical data.

[0138] The conditional probability table represents the probability of a node (behavior) occurring given the parent node.

[0139] Calculate the historical behavior data using maximum likelihood estimation, and its expression is:

[0140]

[0141] Among them, P(X=x|Pa(X)=pa) represents the conditional probability that node X takes the value x under the condition that the value of the parent node is pa, indicating the possibility that the current node occurs when certain conditions of the parent node occur. P represents probability, referring to the possibility that a certain event occurs. X represents a certain node in the network. In a Bayesian network, a node usually represents a certain behavior characteristic of a user. For example, X may represent whether a user purchases product A. X=x represents that node X takes a specific value x. For example, if X represents whether a user purchases product A, then X=yes means the user has purchased product A, and X=no means the user has not purchased product A. Pa(X) represents the set of parent nodes of node X. The parent nodes are those nodes that directly affect the current node. Pa(X)=pa means that the set of parent nodes Pa(X) takes a specific value combination pa. For example, if Pa(X) includes whether the user browses product A, then Pa(X)=yes means the user has browsed product A, and Count represents the number of times a certain event occurs in the historical data.

[0142] For example, X represents whether a user purchases product A. X=yes means the user has purchased product A, and X=no means the user has not purchased product A; Pa(X) represents whether the user browses product A. Pa(X)=yes means the user has browsed product A, and Pa(X)=no means the user has not browsed product A.

[0143] Through historical data, the count is obtained. Count(X=yes, Pa(X)=yes)=58, indicating that the number of times the user has browsed product A and purchased product A is 58. Count(Pa(X)=pa)=100, indicating that the total number of times the user has browsed product A is 100 (including the cases of purchasing and not purchasing after browsing the product).

[0144]

[0145] This means that the probability of the user purchasing product A when browsing product A is 58%.

[0146] Furthermore, by identifying the causal relationships between user behaviors, it is possible to infer future possible behaviors from the user's current behaviors, thus providing more accurate support for recommendations. The causal relationships in the Bayesian network make the recommendation results more interpretable. Through the causal chain, it is possible to explain why a certain product or content is recommended, thereby enhancing user trust.

[0147] S4. Behavioral feature vector A extracted u including but not limited to browsing frequency, click behavior, purchase history, search habits, and dwell time.

[0148] The browsing frequency refers to the browsing frequency of users for different types of goods;

[0149] The click behavior refers to the number of times users click on a certain type of goods or advertisements;

[0150] The purchase history refers to the goods and their categories that users have purchased before;

[0151] The search habits refer to the keywords that users search on the platform;

[0152] The dwell time refers to the dwell time of users on the page of a certain type of goods.

[0153] S4.1. Based on the causal relationship model, output the causal relationship between user behaviors to obtain the possibility of a user purchasing a certain good under the current behavior pattern;

[0154] For example, good C i is a good that the user may purchase, then the possibility of the user purchasing good C under the current behavior pattern i can be expressed as

[0155] S4.2. Generate a personalized recommendation list with priority ranking, and its expression is:

[0156]

[0157] where Q i represents the priority of recommending good i, represents the possibility of purchasing good i under the current user behavior pattern, A u represents the behavioral feature vector of user u, V i represents the weight of good i, V j represents the weight of good j, C j represents the actual interaction behavior of the user purchasing good j, C i represents the actual interaction behavior of the user purchasing good i, and B represents the total number of goods in the current product pool.

[0158] Furthermore, by analyzing the user's behavior pattern and extracting behavioral features, the future needs of users can be accurately predicted, thereby generating a precise personalized recommendation list. Compared with the traditional rule-based recommendation method, the recommendation method based on the causal relationship of user behavior can better capture the interest changes of users and improve the relevance of recommendations.

[0159] S5. Extract the normal behavior range of the user from historical data and set a difference threshold.

[0160] S5.1. To detect changes in real-time behavior, define a time window with a short fixed window to capture the user's behavior data over a certain period of time.

[0161] S5.2. Consider the user's behavior data as time series data, that is, the user's behavior is recorded in chronological order over a certain period of time. To measure the deviation between real-time behavior and historical behavior, use the Euclidean distance to calculate the difference value between real-time behavior data and historical behavior data. The expression is:

[0162]

[0163] where d represents the Euclidean distance between two time series, that is, the difference value between the user's real-time behavior data and historical behavior data, h a represents the a-th data point of the real-time behavior data, y a represents the a-th data point of the historical behavior data, and z represents the length of the time series.

[0164] S5.2.1. After obtaining the difference value, compare this difference value with the previously set difference threshold.

[0165] If the difference value is less than or equal to the difference threshold, it is considered that the current user's behavior conforms to the normal range and no abnormality occurs.

[0166] If the difference value is greater than the difference threshold, it indicates that there is a significant difference between the user's real-time behavior and historical behavior, which may be potential abnormal behavior.

[0167] S5.2.2. Further analyze the reasons for abnormal behavior.

[0168] For example, if the user's purchase behavior suddenly increases significantly, it may be due to the impact of a promotional activity.

[0169] If the user's login location suddenly changes, it may be a sign that the account has been stolen.

[0170] Furthermore, by defining a time window and using a time series difference measurement algorithm, the user's behavior can be monitored in real-time, and abnormal differences from historical behavior data can be quickly detected. This real-time nature can respond immediately when abnormal behavior occurs, reducing potential risks or losses. By setting a difference threshold and combining it with an anomaly detection algorithm, normal behavior fluctuations and real abnormal behavior can be effectively distinguished. Especially in the case of combining historical behavior data, accurate detection can be carried out based on the user's past behavior patterns, reducing the possibility of false alarms and missed detections.

[0171] S6. Divide the differences into frequency differences, category preference differences, spatio-temporal behavior differences, device behavior differences, etc.

[0172] The frequency difference refers to a change in the user's operation frequency.

[0173] For example, if the user suddenly increases the number of product views or purchase frequency, this may reflect a change in the user's interest or the influence of external stimuli (such as promotional activities).

[0174] The category preference difference refers to a change in the user's preference for certain product categories.

[0175] For example, the user mainly browsed electronic products before, but recently began to frequently browse clothing products. This category preference difference may indicate a change in the user's needs.

[0176] The spatio-temporal behavior difference refers to a significant change in the user's access time or location.

[0177] For example, the user usually shops during the day, but recently began to frequently access at night, or the user's login location suddenly changes, which may be a sign of account theft.

[0178] The device behavior difference refers to a change in the user's operating device.

[0179] For example, the user switches from a mobile device to a desktop device, or from a commonly used device to a new device, which may indicate a change in the user's usage scenario and may even involve security issues.

[0180] S6.1. According to the updated user profile, calculate the preference values of the user for different products, reorder the products in the recommendation list, and also consider the influence of the difference type.

[0181] For example, when it is detected that the user's category preference has changed significantly, products related to the new category will be preferentially recommended.

[0182] Furthermore, by dynamically updating the user profile and adjusting the recommendation list in real time, the accuracy of the recommendation can be significantly improved, avoiding the recommendation of outdated products or irrelevant content. Users can see recommended products highly relevant to their current needs and interests on the platform, thereby enhancing user satisfaction and stickiness.

[0183] S6.2. Update the relationships in the knowledge graph according to the user's real-time behavior.

[0184] When the user's preference changes, the relationship weights between the user and nodes such as product categories and brands will be adjusted. For example, if the user frequently purchases products of a certain brand recently, the connection between the user and the brand node will be strengthened.

[0185] If an increase in a user's preference for certain products or categories is detected, adjust the relationship weights between these products and other related products in the knowledge graph. This update can help introduce more relevant products during recommendation, thereby enhancing the diversity and relevance of recommendations.

[0186] S6.3. When significant changes occur in the user's real-time behavior, update the causal relationship model.

[0187] Continuously update the user's behavior pattern through real-time behavior data.

[0188] For example, if the user has recently started frequently purchasing products in a new category, relearn the causal relationship between products in this category and other products, and adjust the recommendation strategy.

[0189] Through the causal relationship model, identify which behaviors have the greatest impact on the user's future purchase decisions, and adjust the recommendation priorities accordingly.

[0190] Furthermore, by updating the cross-domain knowledge graph, a more comprehensive understanding of the user's preferences and needs can be achieved, enhancing the diversity and relevance of product recommendations. The update of the causal relationship model can better capture the causal logic behind the user's behavior, thereby enhancing the rationality and accuracy of the recommendation strategy. Through the update of the knowledge graph and the causal relationship model, some long-tail products (i.e., products that users may be interested in but do not often browse) can be recommended more flexibly, thereby increasing the overall sales volume of the platform.

[0191] This embodiment also provides a personalized recognition and recommendation system based on a knowledge graph, including:

[0192] A data collection module that collects the user's historical behavior data as the basic data for subsequent analysis;

[0193] A preprocessing module that cleans, denoises, formats, and normalizes the collected user behavior data;

[0194] A knowledge graph construction module that constructs a cross-domain knowledge graph based on the preprocessed historical behavior data to capture entity and their association relationships;

[0195] A user profile module that combines the user's behavior data with the knowledge graph to generate a comprehensive user profile;

[0196] A user behavior analysis module that deeply analyzes the user's behavior pattern based on the comprehensive user profile, and extracts the rules and preference patterns of the user's behavior;

[0197] A causal model construction module that constructs a causal relationship model between the user's behaviors according to the comprehensive user profile and the analysis results of the behavior pattern, and identifies the causal chain behind the behaviors;

[0198] The personalized recommendation module generates a personalized recommendation list for users based on the causal relationship model of user behavior, matching their current needs and interests;

[0199] The monitoring and anomaly detection module monitors the latest behavior data of users in real time and uses anomaly detection algorithms to analyze the differences between the latest data and historical behavior data;

[0200] The list adjustment module dynamically adjusts the personalized recommendation list according to the results of real-time behavior and anomaly detection;

[0201] The knowledge graph and model optimization module continuously optimizes the domain knowledge graph and causal relationship model according to the latest behavior data of users.

[0202] This embodiment also provides a computer device applicable to the case of the personalized recognition and recommendation method based on the knowledge graph, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the personalized recognition and recommendation method based on the knowledge graph proposed in the above embodiment.

[0203] This computer device can be a terminal. This computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of this computer device is used to provide computing and control capabilities. The memory of this computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The communication interface of this computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of this computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of this computer device can be a touch layer covered on the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0204] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the personalized recognition and recommendation method based on a knowledge graph as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, abbreviated as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, abbreviated as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, abbreviated as EPROM), programmable read-only memory (Programmable Red-Only Memory, abbreviated as PROM), read-only memory (Read-Only Memory, abbreviated as ROM), magnetic memory, flash memory, magnetic disk or optical disc.

[0205] In summary, the present invention: constructs a cross-domain knowledge graph based on data in multiple domains and generates a comprehensive user profile, realizing a comprehensive understanding of the multi-dimensional and multi-domain behaviors of users. Integrates and analyzes the behavioral characteristics of users in different domains, endows the user profile with semantically rich entity and relationship descriptions, and significantly improves the personalization level of the recommendation function. By dynamically adjusting the recommendation list, the recommendation results can be updated in a timely manner when the user's behavior changes, thus ensuring the real-time and accuracy of the recommendation. By introducing a causal relationship model, it can not only meet the current needs of users, but also identify potential interest points of users, provide personalized recommendation results, greatly improve the user experience and increase the rationality and accuracy of the recommendation.

[0206] Embodiment 2, referring to Table 1, is the second embodiment of the present invention. To further verify the technical solution of the present invention, experimental simulation data of the personalized recognition and recommendation method based on a knowledge graph are given.

[0207] To demonstrate the advantages of the personalized recognition and recommendation method based on the knowledge graph, this embodiment compares this method with existing recommendation methods. The existing recommendation method uses the collaborative filtering algorithm (Collaborative Filtering), which is one of the currently widely used recommendation algorithms. The collaborative filtering algorithm analyzes the historical behavior data of a large number of users and uses the behavior patterns of similar users (such as purchases, browsing, ratings, etc.) to predict the content that the current user may be interested in. The collaborative filtering algorithm is mainly divided into two categories: user-based collaborative filtering and item-based collaborative filtering. In this experiment, the existing recommendation method uses user-based collaborative filtering, that is, by calculating the similarity between users, finding other users with behavior similar to the target user, and then generating a recommendation list for the target user based on the behavior of these similar users.

[0208] The data in this embodiment comes from an e-commerce platform and includes the historical behavior data of 1,000 users in different fields. The data types include click, browse, purchase, and search records. The data collection time span is 6 months, covering multiple fields such as electronic products, clothing, and daily necessities. The existing recommendation method uses the collaborative filtering algorithm to generate a recommendation list for users, while the recommendation method based on the knowledge graph uses a recommendation method that combines a cross-domain knowledge graph and a causal relationship model.

[0209] The implementation steps are as follows:

[0210] Data collection and preprocessing: Collect the historical behavior data of users in the e-commerce platform, including behaviors such as clicks, browsing, purchases, and searches. The data is cleaned, de-duplicated, formatted, and the unstructured data is standardized using natural language processing technology.

[0211] Construction of the cross-domain knowledge graph: Based on the preprocessed data, use entity recognition technology to identify entities in different fields from the user's behavior data, such as the products purchased by the user, the pages browsed, etc. By identifying the explicit and implicit relationships between entities, construct a knowledge graph within the domain, and then through entity alignment and relationship mapping, integrate the knowledge graphs of multiple fields into a unified cross-domain knowledge graph.

[0212] Generation of the user's comprehensive profile: Based on the cross-domain knowledge graph, use a recursive neural network and multi-layer graph embedding technology to generate a dynamic comprehensive profile of the user. Whenever a user generates a new behavior, recursively update the user's profile to capture the evolving characteristics of the user's behavior.

[0213] Behavior Pattern Analysis and Causal Relationship Model Construction: Analyze the cross - domain behavior patterns of users through the Apriori algorithm to identify the associations between behaviors. Then, based on the comprehensive user profile and the results of behavior pattern analysis, construct a causal relationship model to identify the potential causal relationships between user behaviors. Use Bayesian networks to build the causal relationship chain, calculate the conditional probability of each node based on data - driven methods, and finally generate the causal relationship model of user behaviors.

[0214] Personalized Recommendation List Generation: Based on the causal relationship model and the results of behavior pattern analysis, generate a personalized recommendation list for users. Each recommended product is prioritized according to the user behavior feature vector and product weight to ensure that the recommendation list can more accurately reflect the potential needs of users.

[0215] Difference Detection and Real - time Adjustment: Define a real - time behavior window (1 week), and use the time - series difference measurement algorithm to calculate the difference between the user's real - time behavior data and historical behavior data. Based on the difference results, dynamically adjust the recommendation list and optimize the cross - domain knowledge graph and causal relationship model.

[0216] Specifically, it is shown in Table 1 as follows:

[0217] Table 1 Comparison Table of Personalized Recommendation Methods

[0218]

[0219] By comparing the existing collaborative filtering recommendation method with the personalized recommendation method based on the knowledge graph, the data in the table clearly shows the significant differences between the two.

[0220] First of all, from the perspective of the number of user clicks, the recommendation method based on the knowledge graph can better capture the user's interest points. The number of user clicks has increased from 135 times in the collaborative filtering method to 180 times, and the difference value is 45, far exceeding the set difference threshold of 30. This indicates that the recommendation method based on the knowledge graph has higher recommendation relevance, can recommend products that better meet the user's needs, and thus attract user clicks.

[0221] The increase in the number of user purchases is even more significant, from 18 times in the collaborative filtering method to 42 times, and the difference value is 24, exceeding the threshold of 20. This result shows that the recommendation method based on the knowledge graph can not only attract user clicks but also effectively promote conversions, helping users find products that better meet their needs. This difference is due to the fact that this method identifies the deep - seated needs behind user behaviors through the causal relationship model, while the collaborative filtering method only relies on the similarity between users and ignores the causal associations between individual user behaviors.

[0222] The user's stay duration has also increased, from 23 minutes to 30 minutes, with a difference value of 7 minutes, approaching the threshold of 10 minutes. Although the difference value of the stay duration is relatively small, this indicates that the recommendation method based on the knowledge graph can provide more attractive recommended content for users, enabling them to spend more time on the platform. This result further proves the advantage of this method in accurately grasping the user's interest points.

[0223] In terms of the satisfaction of the user recommendation list, the recommendation method based on the knowledge graph performs particularly outstandingly. The user satisfaction has increased from 4.0 points to 4.9 points, with a difference value of 0.9, exceeding the threshold of 0.8. This reflects a significant improvement in the user's recognition of the recommendation list. The recommendation list based on the knowledge graph can more accurately reflect the user's current needs and interests, while the recommendation results of the collaborative filtering method are relatively more generalized.

[0224] In terms of the recognition rate of user abnormal behaviors, the recognition rate of the recommendation method based on the knowledge graph reaches 94%, while that of the collaborative filtering method is only 60%, with a difference value of 34, far exceeding the threshold of 15. This shows that the knowledge graph recommendation method performs more excellently in real-time monitoring of user behaviors and detecting abnormalities. Through the differential analysis of user behaviors, the recommendation strategy can be adjusted in a timely manner to ensure the accuracy and real-time nature of the recommendation results, while the collaborative filtering method is relatively slow in dealing with behavior changes and is difficult to respond quickly.

[0225] Generally speaking, the personalized recommendation method based on the knowledge graph demonstrates obvious advantages. First of all, in terms of the accuracy of personalized recommendations, this method achieves a recommendation accuracy of 92%, while the recommendation accuracy of the collaborative filtering method is only about 70%. Secondly, the knowledge graph method can generate dynamic user portraits, capture the changes in user behaviors in real-time, and predict the user's future needs through a causal relationship model, thereby generating a more targeted recommendation list. In contrast, the collaborative filtering method relies on the similarity between users and cannot fully explore the deep-level associations of individual user behaviors, resulting in generalized and less accurate recommended content.

[0226] In addition, the knowledge graph method performs particularly outstandingly in abnormal behavior recognition and real-time adjustment. By using the time series difference measurement algorithm, it can accurately identify the changes in user behaviors and adjust the recommendation list and optimize the knowledge graph in a timely manner. This dynamic adjustment ability greatly enhances the flexibility of the recommendation method in dealing with complex user behaviors, while the collaborative filtering method lacks this real-time adjustment mechanism and is difficult to cope with the immediate changes in user behaviors.

[0227] In summary, the personalized recognition and recommendation method based on the knowledge graph significantly improves the recommendation accuracy, user satisfaction, and abnormal behavior recognition ability through more accurate user portraits, behavior pattern analysis, and causal relationship models. Especially when dealing with changes in user behavior, it shows higher flexibility and response speed. These innovative and novel effects have been fully verified in the experimental data, demonstrating the uniqueness and advantages of this method in the existing recommendation technologies.

[0228] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A personalized recognition and recommendation method based on a knowledge graph, characterized in that: including, collecting the historical behavior data of users and preprocessing it; constructing a cross-domain knowledge graph based on the preprocessed historical behavior data, generating a comprehensive user portrait, further analyzing the comprehensive user portrait, and obtaining the analysis result of the behavior pattern; constructing a causal relationship model according to the comprehensive user portrait and the analysis result of the behavior pattern, and outputting the causal relationship between user behaviors. Specifically, it includes the following steps: extracting the key behavior features related to potential causal relationships from the comprehensive user portrait; using the key behavior features as the nodes of the Bayesian network; determining the causal relationships between the nodes through a data-driven learning method, and regarding the causal relationships between the nodes as the edges of the Bayesian network; using the maximum likelihood estimation to calculate the historical behavior data, obtaining the conditional probability table of each node and its parent node, and completing the construction of the Bayesian network; using the Bayesian network and generating the causal relationship before user behaviors based on the comprehensive user portrait and the analysis result of the behavior pattern; generating a personalized recommendation list for users based on the causal relationship between user behaviors. Specifically, it includes the following steps: extracting the features representing user behaviors from the analysis result of the user behavior pattern to form the behavior feature vector of the user; calculating according to the causal relationship model and the behavior feature vector of the user to obtain the possibility of purchasing goods under the current behavior pattern of the user; obtaining the future needs of the user based on the causal relationship between user behaviors, and generating a personalized recommendation list with priority ranking, and its expression is: ； Among them, represents the priority of the recommended product ; represents the possibility of purchasing a product in the current behavior pattern of the user ; represents the user 's behavior feature vector ; represents the weight of the product ; represents the weight of the product ; represents the actual interaction behavior of the user when purchasing the product ; represents the actual interaction behavior of the user when purchasing the product represents the total number of products in the current product pool; monitoring user behaviors in real time, and using an anomaly detection algorithm to analyze the difference between the latest behavior data of the user and the historical behavior data of the user; adjusting the personalized recommendation list according to the difference result, and optimizing the cross-domain knowledge graph and the causal relationship model according to the behaviors of the user.

2. The personalized recognition and recommendation method based on a knowledge graph according to claim 1, wherein: collecting the historical behavior data of users and preprocessing it. Specifically, it includes the following steps: collecting the click, browsing, purchase, search, social interaction and health data of users from multiple data sources; cleaning and deduplicating the collected data; processing the unstructured data using natural language processing technology; performing standardization processing on all data.

3. The personalized recognition and recommendation method based on a knowledge graph according to claim 2, wherein: constructing a cross-domain knowledge graph based on the preprocessed historical behavior data, generating a comprehensive user portrait, further analyzing the comprehensive user portrait, and obtaining the analysis result of the behavior pattern. Specifically, it includes the following steps: using entity recognition technology to identify entities from the preprocessed data, and identifying the explicit and implicit relationships between the entities; constructing an independent knowledge graph for each domain respectively, using the identified entities as nodes and the relationships between the entities as edges; integrating the knowledge graphs of each domain into a unified cross-domain knowledge graph through entity alignment and relationship mapping technology; identifying the relationships between the user nodes and other nodes based on the user nodes in the cross-domain knowledge graph, and obtaining the behavior data of the user; analyzing the relationship strength between the user and the entity nodes in different domains according to the behavior data of the user, introducing a recursive neural network and a multi-layer graph embedding technology, recursively generating a new graph layer every time the user performs an operation, capturing the evolution of the user behavior, and generating a dynamic comprehensive portrait of the user; extracting the behavior features in different domains based on the dynamic comprehensive portrait; Analyze the behaviors of users in different fields through the Apriori algorithm and find the associations between these behaviors; Combine and analyze the behavioral characteristics of users in different fields and the associations between behaviors to obtain the results of behavioral analysis.

4. The personalized recognition and recommendation method based on a knowledge graph according to claim 3, wherein: Monitor the user behaviors in real time and use anomaly detection algorithms to analyze the differences between the latest behavioral data of users and their historical behavioral data. The specific steps are as follows. Set a data difference threshold based on historical behavioral data; Define the time window of real-time behaviors; Use the difference measurement algorithm of time series to calculate the difference value between real-time behavioral data and historical behavioral data and compare it with the difference threshold; When the difference value is greater than the difference threshold, analyze these differences, mark the abnormal behaviors, and further analyze the reasons for the abnormal behaviors.

5. The personalized recognition and recommendation method based on a knowledge graph according to claim 4, characterized in that: Adjust the personalized recommendation list according to the difference results, and optimize the cross-domain knowledge graph and causal relationship model according to the behaviors of users. The specific steps are as follows. Identify the types of differences between user behaviors and historical behaviors according to the difference detection results; Adjust the personalized recommendation list based on the dynamically updated user behavior portraits and the types of differences; Adjust and update the cross-domain knowledge graph and causal relationship model according to the real-time behavioral data.

6. A personalized recognition and recommendation system based on a knowledge graph, based on the personalized recognition and recommendation method based on a knowledge graph according to any one of claims 1 to 5, characterized in that: Including, A data collection module that collects the historical behavioral data of users as the basic data for subsequent analysis; A preprocessing module that cleans, denoises, formats, and normalizes the collected user behavioral data; A knowledge graph construction module that constructs a cross-domain knowledge graph based on the preprocessed historical behavioral data to capture entities and their association relationships; A user portrait module that combines the behavioral data of users with the knowledge graph to generate a comprehensive portrait of users; A user behavior analysis module that deeply analyzes the behavior patterns of users based on the comprehensive portraits of users, and extracts the rules and preference patterns of user behaviors; A causal model construction module that constructs a causal relationship model between user behaviors according to the comprehensive portraits of users and the results of behavior pattern analysis, and identifies the causal chain behind the behaviors; A personalized recommendation module that generates a personalized recommendation list for users based on the causal relationship model of user behaviors to match their current needs and interests; A monitoring and anomaly detection module that monitors the latest behavioral data of users in real time and uses anomaly detection algorithms to analyze the differences between the latest data and historical behavioral data; A list adjustment module that dynamically adjusts the personalized recommendation list according to the results of real-time behaviors and anomaly detection; A knowledge graph and model optimization module that continuously optimizes the domain knowledge graph and causal relationship model according to the latest behavioral data of users.

7. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the personalized recognition and recommendation method based on the knowledge graph according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the personalized recognition and recommendation method based on the knowledge graph according to any one of claims 1 to 5.

Images