Object-based general intelligent search system
Through a general intelligent search system based on objects, the business search view model is dynamically adjusted using similar user analysis and clustering algorithms, which solves the problem that traditional systems cannot meet personalized needs and improves search efficiency and user experience.
Patent Information
- Application Number
- CN202411271779.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-11
- Publication Date
- 2024-12-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional business search systems cannot meet the personalized needs of different users, resulting in reduced search efficiency and user experience, especially when facing the differences in users' historical behavioral data, usage habits and business scenarios.
Design a general intelligent search system based on objects, including similar user analysis module, cluster analysis module, user division module, data analysis module and evaluation module, and determine alternative business search view models and Evaluate the ultimate suitable view model and dynamically adjust the search results.
It improves the fine-grainedness of user division, determines the most suitable business search view model, and improves the intelligence and accuracy of the user experience and system.
Smart Images

Figure CN119089024A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligent search, and more specifically, to an object-based general intelligent search system. Background Art
[0002] With the rapid development of information technology, the user needs of enterprise systems are becoming increasingly complex and diverse, especially in business search and query. Traditional business search systems usually meet user needs based on predefined search view models. However, this approach is incapable of meeting the personalized needs of different users. The differences in user historical behavior data, usage habits, and business scenarios often result in the same business search view model being unable to meet the needs of all users, reducing the system's search efficiency and user experience.
[0003] In order to solve the above defects, a technical solution is now provided. Summary of the invention
[0004] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an object-based general intelligent search system to solve the problems raised in the above-mentioned background technology.
[0005] To achieve the above object, the present invention provides the following technical solutions:
[0006] A general intelligent search system based on an object, specifically comprising a similar user analysis module, a cluster analysis module, a user division module, a data analysis module and an evaluation module, with signal connections between the modules;
[0007] A similar user analysis module is used to filter out users who are similar to the current search user and other users in the system by using the user feature business entities of the current search user and other users as features and expressing them in vector form;
[0008] The cluster analysis module is used to perform cluster analysis based on the number of times users use different business search view models, obtain all users in the cluster where the current search user is located according to the cluster where the current search user is located, and mark all users in the cluster where the current search user is located as users of the same type;
[0009] The user classification module is used to determine the users who are commonly marked as the same type of users and similar users by comparing the same type of users with similar users, and obtain the alternative business search view model of the current search user;
[0010] The data analysis module is used to construct a graph structure about the business search view, based on the business entities in the search content of the current search user, to collect the structural information and usage information about the business entity nodes in the search content of the current search user;
[0011] The evaluation module is used to evaluate the relevance of the candidate business search view models to the current search user and determine the business search view model that is ultimately suitable for the current search user.
[0012] In a preferred embodiment, screening similar users of the current search user and other users in the system includes:
[0013] A monitoring interval is set, search history records of the user within the monitoring interval are obtained, business entities in the search history records of the user are extracted, a weight threshold is set, and the weights of different business entities are obtained, wherein the number of each business entity in the search history records of the user within the monitoring interval is obtained, and the total number of user searches in the search history records of the user within the monitoring interval is obtained, and the ratio of the number of each business entity in the search history records of the user within the monitoring interval to the total number of user searches in the search history records of the user within the monitoring interval is used as the weight of different business entities, the weights of different business entities of the user are compared with the weight threshold, business entities greater than the weight threshold are obtained, the number of business entity characteristic values is set, business entities of the user greater than the weight threshold are sorted from large to small according to the size of the weight, and business entities with the number of business entity characteristic values are extracted as user characteristic business entities;
[0014] Perform similarity analysis on the current search user and other users. Use the user feature business entities of the current search user and other users as features to represent them in vector form. Use the similarity calculation method, set a similarity threshold, determine users whose similarity with other users is greater than the similarity threshold, and regard users whose similarity with other users is greater than the similarity threshold as similar users.
[0015] In a preferred embodiment, marking all users in the cluster where the user belongs as the same type of users includes:
[0016] The K-means clustering algorithm is used to perform cluster analysis on the users in the system. The unit time length is set, and cluster analysis is performed based on the number of times the user uses different business search view models in the unit time length. The number of times the user uses different business search view models in the unit time length is represented by a vector, that is, YH n =[CS1, CS2, CS3,…,CS i ], where n = 1, 2, 3, ..., N, N is a positive integer, n is the user's number, i = 1, 2, 3, ..., I, I is a positive integer, i is the number of the business search view model, YH n is the user’s feature vector;
[0017] Randomly select K users and divide them into K groups, and use the feature vectors of the randomly selected K users as the initial cluster centers;
[0018] Calculate the Euclidean distance between the user and the cluster center, and assign the user to the cluster with the nearest cluster center;
[0019] Calculate the mean of the user feature vectors in each cluster as the new cluster center;
[0020] Repeat the above steps until the location of the cluster center no longer changes significantly or the set number of iterations is reached;
[0021] After clustering, K clusters are obtained, and the K clusters are marked to obtain a set of user types;
[0022] According to the cluster where the current search user is located, all users in the cluster where the current search user is located are obtained, and all users in the cluster where the current search user is located are marked as the same type of users.
[0023] In a preferred implementation, obtaining the candidate business search view model of the current search user includes:
[0024] By comparing the same type of users with similar users, users who are commonly marked as the same type of users and similar users are determined, and the users who are commonly marked as the same type of users and similar users are recorded as class cross-identification users. The business search view model of the class cross-identification users is collected, and the business search view model of the class cross-identification users is used as the alternative business search view model of the current search user.
[0025] In a preferred embodiment, collecting structural information about business entity nodes in the current search user's search content includes:
[0026] The structural information of the business entity node is represented by the structural centrality deviation coefficient. The logic of obtaining the structural centrality deviation coefficient is as follows: obtain the business entity extracted from the user's search content, determine the relative importance of the business entity node using the PageRank algorithm according to the graph structure of the candidate business search view model, initialize the PageRank value, set the initial PageRank values of all nodes to be equal, and use the PageRank formula for iterative update. The formula is expressed as: Among them, PR u is the relative importance of business entity node u, PR v is the relative importance of business entity node v, d is the damping coefficient, N is the total number of business entity nodes, B u Represents the set of business entity nodes pointing to business entity node u, L v is the out-degree of the business entity node v;
[0027] Obtain the relative importance of extracting the business entity from the user's search content, and mark the relative importance of extracting the business entity from the user's search content as: ZY m , where m=1, 2, 3, ..., M, M is a positive integer, and m is the number of the business entity extracted from the user's search content;
[0028] Calculate the relative importance coefficient of the user in the alternative business search view model. The calculation formula is: Among them, ZY all is the relative importance coefficient;
[0029] According to the graph structure of the alternative business search view model, the proximity centrality of the business entity extracted from the user's search content is obtained, the inverse of the distance from the business entity node to all other nodes is obtained, and the inverse of the distance from the business entity node to all other nodes is used as the proximity centrality of the business entity node, and the proximity centrality of the business entity extracted from the user's search content is marked as: JJ m ;
[0030] Calculate the closeness centrality coefficient of the user in the alternative business search view model. The calculation formula is: Among them, JJ all is the closeness centrality coefficient;
[0031] According to the graph structure of the alternative business search view model, the betweenness centrality of the business entities extracted from the user's search content is obtained, the shortest distance of each pair of business entity nodes in the graph structure is obtained, the number of paths that each node in the graph structure passes through the business entity node in all shortest paths is obtained, and the number of paths that each node in the graph structure passes through the business entity node in all shortest paths is marked as: ZD m , obtain the total number of paths in the graph structure, and mark the total number of paths in the graph structure as: LJ. The calculation formula for the betweenness centrality of the business entity extracted from the user's search content is:
[0032] Calculate the betweenness centrality coefficient of the user in the alternative business search view model. The calculation formula is: Among them, JS all is the betweenness centrality coefficient;
[0033] Set the standard value of the relative importance coefficient, the standard value of the closeness centrality coefficient, and the standard value of the betweenness centrality coefficient, obtain the deviations of the standard value of the relative importance coefficient, the standard value of the closeness centrality coefficient, and the standard value of the betweenness centrality coefficient from the relative importance coefficient, the closeness centrality coefficient, and the betweenness centrality coefficient, and mark the deviation of the relative importance coefficient, the closeness centrality coefficient, and the betweenness centrality coefficient as: PC zY 、PCJJ and PC JS ;
[0034] Calculate the structural centrality deviation coefficient, the calculation formula is: Among them, PC jg is the structural centrality degree deviation coefficient.
[0035] In a preferred embodiment, collecting usage information about the business entity nodes in the current search user's search content includes:
[0036] The usage information of the business entity node is represented by a relative application weight coefficient, and the acquisition logic of the relative application weight coefficient is: obtaining the number of times the business entity extracted from the user's search content uses different candidate business search view models, and marking the number of times the business entity extracted from the user's search content uses different candidate business search view models as: Wherein, g=1, 2, 3, ..., G, G is a positive integer, and g is the serial number of the candidate business search view model;
[0037] It should be noted that different users will use different business search view models during the search process. By counting the historical records of the system, it is possible to obtain in which business search view models the business entity is used more frequently.
[0038] The standard deviation of the number of times the business entity uses different alternative business search view models is obtained, and the standard deviation of the number of times the business entity uses different alternative business search view models is marked as: σ m ,in,
[0039] Calculate the relative application weight coefficient, the calculation formula is: Among them, YY g Relative application weight coefficient of the view model g for the candidate business search.
[0040] In a preferred embodiment, the business search view model that is finally suitable for the current search user includes:
[0041] The structural centrality degree deviation coefficient and the relative application weight coefficient are calculated by weighted summation to obtain the alternative business search view model evaluation coefficient;
[0042] Setting a candidate business search view model evaluation coefficient threshold, and comparing the candidate business search view model evaluation coefficient with the candidate business search view model evaluation coefficient threshold;
[0043] If the candidate business search view model evaluation coefficient is greater than the candidate business search view model evaluation coefficient threshold, the candidate business search view model is used as the business search view model required by the current search user;
[0044] If there are multiple business search view models required by the current search user, the alternative business search view model with a larger alternative business search view model evaluation coefficient is preferentially selected as the business search view model required by the current search user according to the size of the alternative business search view model evaluation coefficient.
[0045] Technical effects and advantages of the present invention:
[0046] The present invention screens similar users through the users' historical search contents, and divides the users who use the same business search view model into corresponding groups by using a clustering algorithm, and finally screens out the cross-class identification users by combining the two methods, thereby improving the granularity of user division, and determining the business search view model that the current search user may choose to use. The present invention extracts the key information of the user's current search through NLP technology, and finally determines the business search view model that best suits the current search user by combining the structural information and usage information between the business search view models. The present invention helps to improve the user experience and enhance the intelligence and accuracy of the system. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings;
[0048] Figure 1 It is a structural schematic diagram of an object-based general intelligent search system of the present invention. DETAILED DESCRIPTION
[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0050] Example 1
[0051] Figure 1 A schematic diagram of the structure of a general intelligent search system based on an object is given, which specifically includes a similar user analysis module, a cluster analysis module, a user division module, a data analysis module and an evaluation module, and signal connections between the modules;
[0052] A similar user analysis module is used to filter out users who are similar to the current search user and other users in the system by using the user feature business entities of the current search user and other users as features and expressing them in vector form;
[0053] The cluster analysis module is used to perform cluster analysis based on the number of times users use different business search view models, obtain all users in the cluster where the current search user is located according to the cluster where the current search user is located, and mark all users in the cluster where the current search user is located as users of the same type;
[0054] The user classification module is used to determine the users who are commonly marked as the same type of users and similar users by comparing the same type of users with similar users, and obtain the alternative business search view model of the current search user;
[0055] The data analysis module is used to construct a graph structure about the business search view, based on the business entities in the search content of the current search user, to collect the structural information and usage information about the business entity nodes in the search content of the current search user;
[0056] The evaluation module is used to evaluate the relevance of the candidate business search view models to the current search user and determine the business search view model that is ultimately suitable for the current search user.
[0057] The object-based universal intelligent search system is a flexible and intelligent system architecture that supports query and search requirements in various business scenarios. It extracts user input and user permissions, dynamically adjusts the business search view model, and provides users with the most matching search results. The business search view model of the system includes business entities, business entity metadata, business search view models, and business indicators.
[0058] Business entity: an abstract description of the subject of a business scenario. Its basic attributes include: entity unique identifier, name, result set description, type, period, etc.
[0059] Business entity metadata: an abstract description of all specific data associated with the subject of a business scenario. Basic attributes include: unique identifier, associated entity identifier, basic data type, extended data type, data length, presentation sequence number, etc. A metadata must belong to a business entity, and a business entity is associated with at least one or more metadata.
[0060] As a storage layer, business entities and metadata are usually bound to specific data tables or files in a specific data storage engine. The data storage engine can be flexibly replaced while the basic definitions of entities and metadata remain unchanged.
[0061] Business search view: a description of user query and search function requirements. Basic attributes include: unique identifier, view name, view permissions, retrieval rule configuration, etc.
[0062] Business indicator: a description of the specific display needs of user query and search function requirements. The basic attributes include: unique identifier, associated view identifier, indicator name, parent indicator identifier, associated metadata identifier, whether to be used as a search condition, whether to be used as a list header, data conversion rules, front-end interaction and UI display rules, display sequence number, etc.
[0063] Business indicators can be defined at multiple levels and are dependent on each other to form a business indicator tree. The last-level business indicator must be bound to a business entity metadata to determine the business data description associated with this indicator. By defining a business search view and using at least one or more business indicators in the view, the display, search, conversion and other functional definitions of the query search list are completed.
[0064] The core components of the system are used to call and manage business search view models, and collaborate with each other through different responsibilities and functions to achieve efficient search functions and user request processing. The system follows the layered design principle, and the core components include the interface layer, rule base, interpretation layer, and execution layer. The specific explanations are as follows:
[0065] Interface layer: responsible for receiving view parameters, retrieval and permission rules, UI parameters, indicator parameters and other information, interacting with front-end users, submitting various parameters after user interaction to the back-end, and scheduling all logic as needed;
[0066] Rule base: Aggregates all business search view models, stores all pre-defined views, indicators, entities, metadata and other information, and provides various services based on cache to complete client view rendering, indicator parameter conversion query search conditions, etc.;
[0067] Interpretation layer: responsible for interpreting the query search conditions into a complete query statement of a database or search engine, and providing a variable pool to store various built-in parameters;
[0068] Execution layer: responsible for sending query statements to the database or search engine to execute the query, returning the result set, and visually configuring data desensitization and permission filtering.
[0069] The business search view model is created and configured by managers, who need to regularly maintain and update the view model to adapt to changes in business needs or system upgrades, including adding new fields, adjusting layouts, modifying retrieval rules, etc. Managers are responsible for setting and managing the permission control of the view model to ensure that only authorized users can access or modify specific view models. When user needs change, managers can adjust the configuration of the view model according to the new needs and store the business search view model for the convenience of subsequent user calls.
[0070] A monitoring interval is set, search history records within the user monitoring interval are obtained, business entities in the user search history records are extracted, a weight threshold is set, and the weights of different business entities are obtained, wherein the number of each business entity in the user search history records within the monitoring interval and the total number of user searches in the user search history records within the monitoring interval are obtained, and the ratio of the number of each business entity in the user search history records within the monitoring interval to the total number of user searches in the user search history records within the monitoring interval is used as the weight of different business entities, the weights of different business entities of the user are compared with the weight threshold, business entities greater than the weight threshold are obtained, the number of business entity characteristic values is set, business entities of the user greater than the weight threshold are sorted from large to small according to the size of the weight, and business entities with the number of business entity characteristic values are extracted as user characteristic business entities.
[0071] It should be noted that the monitoring interval is a specific time period that can reflect the user's actual search status. The weight threshold is a specific value that is used to reflect the user's interest points in the search history. The number of business entity feature values is a specific value that is used to reflect the business entities that the user is most interested in. The monitoring interval, weight threshold, and number of business entity feature values are set by professional staff and are not elaborated here. If the weights of different business entities of the user are all less than the weight threshold or the number of business entities of the user greater than the weight threshold is less than the number of business entity feature values, the user will not be analyzed.
[0072] Perform similarity analysis on the current search user and other users. Use the user feature business entities of the current search user and other users as features to represent them in vector form. Use the similarity calculation method, set a similarity threshold, determine users whose similarity with other users is greater than the similarity threshold, and regard users whose similarity with other users is greater than the similarity threshold as similar users.
[0073] It should be noted that common similarity calculation methods include Euclidean distance, Hamming distance, and cosine similarity. An appropriate similarity calculation method is selected to determine similar users to the current search user. The similarity threshold is set by professional staff.
[0074] The K-means clustering algorithm is used to perform cluster analysis on the users in the system. The unit time length is set, and cluster analysis is performed based on the number of times the user uses different business search view models in the unit time length. The number of times the user uses different business search view models in the unit time length is represented by a vector, that is, YH n =[CS1, CS2, CS3,…,CS i], where n = 1, 2, 3, ..., N, N is a positive integer, n is the user's number, i = 1, 2, 3, ..., I, I is a positive integer, i is the number of the business search view model, YH n is the user’s feature vector;
[0075] Randomly select K users and divide them into K groups, and use the feature vectors of the randomly selected K users as the initial cluster centers;
[0076] Calculate the Euclidean distance between the user and the cluster center, and assign the user to the cluster with the nearest cluster center;
[0077] Calculate the mean of the user feature vectors in each cluster as the new cluster center;
[0078] Repeat the above steps until the location of the cluster center no longer changes significantly or the set number of iterations is reached;
[0079] The clustering objective function takes the form of a squared error criterion function: Where, J: clustering objective function value (sum of squared errors), k is the number of clusters, is the feature vector of the nth user in cluster k;
[0080] After clustering, K clusters are obtained, and the K clusters are marked to obtain the user type set. The user type set is recorded as: H = {C1, C2, C3, ..., C K}.
[0081] For example, suppose there is an internal enterprise system where users can query data through different business search view models. Each user may use multiple business search view models, but their usage habits may be different.
[0082] There are five business search view models in the enterprise internal system, namely: view model A, view model B, view model C, view model D and view model E. The number of times users use each view model in a unit time length is counted;
[0083] The feature vectors of users are represented by the number of times they use each view model, that is, [number of times view model A is used, number of times view model B is used, number of times view model C is used, number of times view model D is used, number of times view model E is used]. Using these feature vectors as input, the K-means clustering algorithm is applied to divide the user groups.
[0084] According to the cluster where the current search user is located, all users in the cluster where the current search user is located are obtained, and all users in the cluster where the current search user is located are marked as same-type users. By comparing same-type users with similar users, users who are commonly marked as same-type users and similar users are determined, and users who are commonly marked as same-type users and similar users are recorded as class cross-identified users. Business search view models of class cross-identified users are collected, and the business search view models of class cross-identified users are used as alternative business search view models for the current search user.
[0085] Construct a graph structure for the business search view, where the graph structure of the business search view is composed of business entity nodes, business entity metadata nodes, business search view nodes, and business indicator nodes. The relationship between nodes and edges in the graph structure of the business search view is expressed as follows:
[0086] The edge between a business entity node and a business entity metadata node: represents the subordinate relationship between a business entity and its associated metadata;
[0087] The edge between the business search view node and the business indicator node: indicates which specific business indicators a business search view contains. There is only one business search view in a business search view model.
[0088] The edge between a business indicator node and a business entity metadata node indicates that a certain indicator is defined based on specific business entity data.
[0089] Extract the business entities in the current search user's search content, and based on the business entities in the current search user's search content, collect the structural information and usage information about the business entity node, and represent the structural information of the business entity node through the structural centrality degree deviation coefficient, and represent the usage information of the business entity node through the relative application weight coefficient.
[0090] It should be noted that the business entities extracted from the user's search content using NLP technology, including keyword extraction, named entity recognition (NER), semantic analysis and other technologies, are matched with the extracted keywords or phrases and predefined business entities in the system to identify the business entities of interest to the user.
[0091] The logic for obtaining the structural centrality deviation coefficient is as follows: obtain the business entity extracted from the user's search content, determine the relative importance of the business entity node using the PageRank algorithm according to the graph structure of the candidate business search view model, initialize the PageRank value, set the initial PageRank values of all nodes to be equal, and iterate and update using the PageRank formula, which is expressed as follows: Among them, PR uis the relative importance of business entity node u, PR v is the relative importance of business entity node v, d is the damping coefficient, N is the total number of business entity nodes, B u Represents the set of business entity nodes pointing to business entity node u, L v is the out-degree of the business entity node v;
[0092] Obtain the relative importance of extracting the business entity from the user's search content, and mark the relative importance of extracting the business entity from the user's search content as: ZY m , where m=1, 2, 3, ..., M, M is a positive integer, and m is the number of the business entity extracted from the user's search content;
[0093] Calculate the relative importance coefficient of the user in the alternative business search view model. The calculation formula is: Among them, ZY all is the relative importance coefficient;
[0094] According to the graph structure of the alternative business search view model, the proximity centrality of the business entity extracted from the user's search content is obtained, the inverse of the distance from the business entity node to all other nodes is obtained, and the inverse of the distance from the business entity node to all other nodes is used as the proximity centrality of the business entity node, and the proximity centrality of the business entity extracted from the user's search content is marked as: JJ m ;
[0095] Calculate the closeness centrality coefficient of the user in the alternative business search view model. The calculation formula is: Among them, JJ all is the closeness centrality coefficient;
[0096] According to the graph structure of the alternative business search view model, the betweenness centrality of the business entities extracted from the user's search content is obtained, and the shortest distance of each pair of business entity nodes in the graph structure is obtained, where the shortest path refers to the path with the least number of edges from one node to another node. The number of paths that each node in the graph structure passes through the business entity node in all the shortest paths is obtained, and the number of paths that each node in the graph structure passes through the business entity node in all the shortest paths is marked as: ZD m , obtain the total number of paths in the graph structure, and mark the total number of paths in the graph structure as: LJ. The calculation formula for the betweenness centrality of the business entity extracted from the user's search content is:
[0097] Calculate the betweenness centrality coefficient of the user in the alternative business search view model. The calculation formula is: Among them, JS allis the betweenness centrality coefficient;
[0098] Set the standard value of the relative importance coefficient, the standard value of the closeness centrality coefficient, and the standard value of the betweenness centrality coefficient, obtain the deviations of the standard value of the relative importance coefficient, the standard value of the closeness centrality coefficient, and the standard value of the betweenness centrality coefficient from the relative importance coefficient, the closeness centrality coefficient, and the betweenness centrality coefficient, and mark the deviation of the relative importance coefficient, the closeness centrality coefficient, and the betweenness centrality coefficient as: PC ZY 、PC JJ and PC JS , where the ratios of the standard values of the relative importance coefficient, the closeness centrality coefficient, and the betweenness centrality coefficient to the relative importance coefficient, the closeness centrality coefficient, and the betweenness centrality coefficient are taken as the relative importance coefficient deviation, the closeness centrality coefficient deviation, and the betweenness centrality coefficient deviation respectively;
[0099] Calculate the structural centrality deviation coefficient, the calculation formula is: Among them, PC jg is the structural centrality degree deviation coefficient.
[0100] It can be seen from the formula that the larger the structural centrality deviation coefficient is, the more important the business entity extracted from the user's search content may be in the graph structure in the alternative business search view model, indicating that it is more appropriate to use the alternative business search view model to answer the user's search content.
[0101] The logic for obtaining the relative application weight coefficient is: obtaining the number of times the business entity extracted from the user's search content uses different candidate business search view models, and marking the number of times the business entity extracted from the user's search content uses different candidate business search view models as: Wherein, g=1, 2, 3, ..., G, G is a positive integer, and g is the serial number of the candidate business search view model;
[0102] It should be noted that different users will use different business search view models during the search process. By counting the historical records of the system, it is possible to obtain in which business search view models the business entity is used more frequently.
[0103] The standard deviation of the number of times the business entity uses different alternative business search view models is obtained, and the standard deviation of the number of times the business entity uses different alternative business search view models is marked as: σ m ,in,
[0104] Calculate the relative application weight coefficient, the calculation formula is: Among them, YY gRelative application weight coefficient of the view model g for the candidate business search.
[0105] The structural centrality degree deviation coefficient and the relative application weight coefficient are calculated by weighted summation to obtain the alternative business search view model evaluation coefficient. The calculation formula of the alternative business search view model evaluation coefficient is: pg g =α1PC jg +α2YY g , where pg g is the alternative business search view model evaluation coefficient of the alternative business search view model g, α1 and α2 are the proportional coefficients of the structural centrality degree deviation coefficient and the relative application weight coefficient, and α1 and α2 are respectively greater than 0.
[0106] Set an alternative business search view model evaluation coefficient threshold, compare the alternative business search view model evaluation coefficient with the alternative business search view model evaluation coefficient threshold, if the alternative business search view model evaluation coefficient is greater than the alternative business search view model evaluation coefficient threshold, then use the alternative business search view model as the business search view model required by the current search user; if the alternative business search view model evaluation coefficient is less than the alternative business search view model evaluation coefficient threshold, then do not use the alternative business search view model as the business search view model required by the current search user; if there are multiple business search view models required by the current search user, then according to the size of the alternative business search view model evaluation coefficient, give priority to selecting the alternative business search view model with a larger alternative business search view model evaluation coefficient as the business search view model required by the current search user; if there is no business search view model required by the current search user, then use the general business search view model, and send a signal to the management personnel, so that the management personnel adjust the business search view model.
[0107] It should be noted that the general business search view model is a business search view model used by new users or users who do not have the ability to use the above method. It is specifically set by staff in professional fields and will not be described in detail here.
[0108] The present invention screens similar users through the users' historical search contents, and divides the users who use the same business search view model into corresponding groups by using a clustering algorithm, and finally screens out the cross-class identification users by combining the two methods, thereby improving the granularity of user division, and determining the business search view model that the current search user may choose to use. The present invention extracts the key information of the user's current search through NLP technology, and finally determines the business search view model that best suits the current search user by combining the structural information and usage information between the business search view models. The present invention helps to improve the user experience and enhance the intelligence and accuracy of the system.
[0109] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.
[0110] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.
[0111] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.
[0112] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.
[0113] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0114] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
[0115] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks or optical disks.
[0116] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.
Claims
1. A general intelligent search system based on objects, characterized in that ,Specifically it includes similar user analysis module, clustering analysis module, user segmentation module, data analysis module and evaluation module, and signal connection between modules; A similar user analysis module is used to filter out users who are similar to the current search user and other users in the system by using the user feature business entities of the current search user and other users as features and expressing them in vector form; The cluster analysis module is used to perform cluster analysis based on the number of times users use different business search view models, obtain all users in the cluster where the current search user is located according to the cluster where the current search user is located, and mark all users in the cluster where the current search user is located as users of the same type; The user classification module is used to determine the users who are commonly marked as the same type of users and similar users by comparing the same type of users with similar users, and obtain the alternative business search view model of the current search user; The data analysis module is used to construct a graph structure about the business search view, based on the business entities in the search content of the current search user, to collect the structural information and usage information about the business entity nodes in the search content of the current search user; The evaluation module is used to evaluate the relevance of the candidate business search view models to the current search user and determine the business search view model that is ultimately suitable for the current search user.
2. According to claim 1, a general intelligent search system based on objects is characterized in that ,Filter similar users of the current search user to other users in the system, including: A monitoring interval is set, search history records of the user within the monitoring interval are obtained, business entities in the search history records of the user are extracted, a weight threshold is set, and the weights of different business entities are obtained, wherein the number of each business entity in the search history records of the user within the monitoring interval is obtained, and the total number of user searches in the search history records of the user within the monitoring interval is obtained, and the ratio of the number of each business entity in the search history records of the user within the monitoring interval to the total number of user searches in the search history records of the user within the monitoring interval is used as the weight of different business entities, the weights of different business entities of the user are compared with the weight threshold, business entities greater than the weight threshold are obtained, the number of business entity characteristic values is set, business entities of the user greater than the weight threshold are sorted from large to small according to the size of the weight, and business entities with the number of business entity characteristic values are extracted as user characteristic business entities; Perform similarity analysis on the current search user and other users. Use the user feature business entities of the current search user and other users as features to represent them in vector form. Use the similarity calculation method, set a similarity threshold, determine users whose similarity with other users is greater than the similarity threshold, and regard users whose similarity with other users is greater than the similarity threshold as similar users.
3. The object-based general intelligent search system according to claim 1, characterized in that: Mark all users in the cluster as the same type of users, including: The K-means clustering algorithm is used to perform cluster analysis on the users in the system. The unit time length is set, and cluster analysis is performed based on the number of times the user uses different business search view models in the unit time length. The number of times the user uses different business search view models in the unit time length is represented by a vector, that is, YH n =[CS1, CS2, CS3,…,CS i ], where n = 1, 2, 3, ..., N, N is a positive integer, n is the user's number, i = 1, 2, 3, ..., I, I is a positive integer, i is the number of the business search view model, YH n is the user’s feature vector; Randomly select K users and divide them into K groups, and use the feature vectors of the randomly selected K users as the initial cluster centers; Calculate the Euclidean distance between the user and the cluster center, and assign the user to the cluster with the nearest cluster center; Calculate the mean of the user feature vectors in each cluster as the new cluster center; Repeat the above steps until the location of the cluster center no longer changes significantly or the set number of iterations is reached; After clustering, K clusters are obtained, and the K clusters are marked to obtain a set of user types; According to the cluster where the current search user is located, all users in the cluster where the current search user is located are obtained, and all users in the cluster where the current search user is located are marked as the same type of users.
4. The object-based general intelligent search system according to claim 1, characterized in that: Get the alternative business search view model of the current search user, including: By comparing the same type of users with similar users, users who are commonly marked as the same type of users and similar users are determined, and the users who are commonly marked as the same type of users and similar users are recorded as class cross-identification users. The business search view model of the class cross-identification users is collected, and the business search view model of the class cross-identification users is used as the alternative business search view model of the current search user.
5. The object-based general intelligent search system according to claim 4, characterized in that: Collect structural information about the business entity nodes in the current search user's search content, including: The structural information of the business entity node is represented by the structural centrality deviation coefficient. The logic for obtaining the structural centrality deviation coefficient is as follows: obtain the business entity extracted from the user's search content, determine the relative importance of the business entity node using the PageRank algorithm according to the graph structure of the candidate business search view model, initialize the PageRank value, set the initial PageRank values of all nodes to be equal, and iterate and update using the PageRank formula, which is expressed as follows: Among them, PR u is the relative importance of business entity node u, PR v is the relative importance of business entity node v, d is the damping coefficient, N is the total number of business entity nodes, B u Represents the set of business entity nodes pointing to business entity node u, L v is the out-degree of the business entity node v; Obtain the relative importance of extracting the business entity from the user's search content, and mark the relative importance of extracting the business entity from the user's search content as: ZY m , where m=1, 2, 3, ..., M, M is a positive integer, and m is the number of the business entity extracted from the user's search content; Calculate the relative importance coefficient of the user in the alternative business search view model. The calculation formula is: Among them, ZY all is the relative importance coefficient; According to the graph structure of the alternative business search view model, the proximity centrality of the business entity extracted from the user's search content is obtained, the inverse of the distance from the business entity node to all other nodes is obtained, and the inverse of the distance from the business entity node to all other nodes is used as the proximity centrality of the business entity node, and the proximity centrality of the business entity extracted from the user's search content is marked as: JJ m ; Calculate the closeness centrality coefficient of the user in the alternative business search view model. The calculation formula is: Among them, JJ all is the closeness centrality coefficient; According to the graph structure of the alternative business search view model, the betweenness centrality of the business entities extracted from the user's search content is obtained, the shortest distance of each pair of business entity nodes in the graph structure is obtained, the number of paths that each node in the graph structure passes through the business entity node in all shortest paths is obtained, and the number of paths that each node in the graph structure passes through the business entity node in all shortest paths is marked as: ZD m , obtain the total number of paths in the graph structure, and mark the total number of paths in the graph structure as: LJ. The calculation formula for the betweenness centrality of the business entity extracted from the user's search content is: Calculate the betweenness centrality coefficient of the user in the alternative business search view model. The calculation formula is: Among them, JS all is the betweenness centrality coefficient; Set the standard value of the relative importance coefficient, the standard value of the closeness centrality coefficient, and the standard value of the betweenness centrality coefficient, obtain the deviations of the standard value of the relative importance coefficient, the standard value of the closeness centrality coefficient, and the standard value of the betweenness centrality coefficient from the relative importance coefficient, the closeness centrality coefficient, and the betweenness centrality coefficient, and mark the deviation of the relative importance coefficient, the closeness centrality coefficient, and the betweenness centrality coefficient as: PC ZY 、PC JJ and PC JS ; Calculate the structural centrality deviation coefficient, the calculation formula is: Among them, PC jg is the structural centrality degree deviation coefficient.
6. The object-based general intelligent search system according to claim 4, characterized in that: Collect usage information about the business entity nodes in the current search user's search content, including: The usage information of the business entity node is represented by a relative application weight coefficient, and the acquisition logic of the relative application weight coefficient is: obtaining the number of times the business entity extracted from the user's search content uses different candidate business search view models, and marking the number of times the business entity extracted from the user's search content uses different candidate business search view models as: Wherein, g=1, 2, 3, ..., G, G is a positive integer, and g is the serial number of the candidate business search view model; It should be noted that different users will use different business search view models during the search process. By counting the historical records of the system, it is possible to obtain in which business search view models the business entity is used more frequently. The standard deviation of the number of times the business entity uses different alternative business search view models is obtained, and the standard deviation of the number of times the business entity uses different alternative business search view models is marked as: σ m ,in, Calculate the relative application weight coefficient, the calculation formula is: Among them, YY g Relative application weight coefficient of the view model g for the candidate business search.
7. The object-based general intelligent search system according to claim 4, characterized in that: The business search view model that is ultimately suitable for the current search user includes: The structural centrality degree deviation coefficient and the relative application weight coefficient are calculated by weighted summation to obtain the alternative business search view model evaluation coefficient; Setting a candidate business search view model evaluation coefficient threshold, and comparing the candidate business search view model evaluation coefficient with the candidate business search view model evaluation coefficient threshold; If the candidate business search view model evaluation coefficient is greater than the candidate business search view model evaluation coefficient threshold, the candidate business search view model is used as the business search view model required by the current search user; If there are multiple business search view models required by the current search user, the alternative business search view model with a larger alternative business search view model evaluation coefficient is preferentially selected as the business search view model required by the current search user according to the size of the alternative business search view model evaluation coefficient.
Citation Information
Patent Citations
Marketing method and system based on customer clustering
CN109615426A
Broadcasting and television technology knowledge recommendation method based on user portrait and knowledge graph
CN115168744A
Business recommendation method and device, equipment and storage medium
CN117541298A
Artificial intelligence data search and distribution method and system
CN118467851A
Ranking Entity Based Search Results Using User Clusters
US20200073953A1
