Metadata-driven book search and recommendation method based on virtual knowledge graph

Abstract

The present application relates to the field of artificial intelligence, in particular to a metadata-driven book search and recommendation method based on a virtual knowledge graph, comprising: receiving a query request, triggering multi-source metadata aggregation, generating a temporary virtual node, establishing an inter-node association edge and marking a relationship weight through semantic vector calculation; the virtual node is automatically released after the query response, the cached metadata is reused for secondary query, the virtual node attributes are classified according to stability and marked with time efficiency; the query is parsed into triples, a simple knowledge organization system vocabulary and citation data are used to construct a citation chain to realize accurate association of cross-source entities; the search engine returns matching results through semantic analysis, sub-graph generation and association strength, and the recommendation engine generates recommendations based on multi-hop path reasoning. The secondary query of the present application reuses cached metadata, significantly improves response speed, effectively reduces server load and avoids resource redundancy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, in particular to a metadata-driven book search and recommendation method based on a virtual knowledge graph. BACKGROUND

[0002] The rapid development of artificial intelligence technology has injected strong impetus into innovation in various industries. However, the current recommendation system of libraries still mainly relies on user behavior data, and book-related information such as authors, publishers, and sales is scattered in various databases, making it difficult to achieve deep correlation. Although some libraries have already developed knowledge graphs, due to the frequent updating of library book information, existing knowledge graphs are difficult to adapt to changes in dynamic metadata. In addition, the lack of semantic understanding ability also leads to low search accuracy. For example, when searching for "quantum physics for beginners", it is difficult to effectively associate with related content such as "quantum mechanics basics". SUMMARY

[0003] The present application provides a metadata-driven book search and recommendation method based on a virtual knowledge graph to solve the technical problems in the prior art.

[0004] The technical solution of the present application to solve the above technical problems is as follows: a metadata-driven book search and recommendation method based on a virtual knowledge graph, comprising the following steps:

[0005] S101, receiving a query request, triggering multi-source metadata aggregation, generating a temporary virtual node, and establishing an associated edge between nodes and marking the relationship weight through semantic vector calculation;

[0006] S102, the virtual node is automatically released after the query response, and the cached metadata is reused for secondary query. The virtual node attributes are classified and labeled according to stability and time efficiency;

[0007] S103, parsing the query into triples, constructing a citation chain with a simple knowledge organization system vocabulary and citation data, and realizing accurate correlation between cross-source entities;

[0008] S104, the search engine returns the matching results through semantic analysis, sub-graph generation, and correlation strength, and the recommendation engine generates recommendations based on multi-hop path reasoning.

[0009] In a preferred embodiment, in S101, the query request input by the user is received, the query request is limited to ISBN, book name and keyword, the parsing unit adopts a differentiated processing mechanism to parse the query request, for the ISBN type, the standard format is matched through a regular expression, for the book name type, a string fuzzy matching algorithm is adopted to process input errors, and for the keyword type, a core retrieval identifier is extracted through a subject word extraction technology, all parsing results are converted into standardized retrieval identifiers, and based on the standardized retrieval identifiers, a distributed metadata collector is triggered to perform real-time aggregation, the data sources are divided into three categories, including a basic data source, a dynamic data source and an associated data source, the metadata collector is connected to each data source through a standardized API interface to ensure uniform data format;

[0010] A temporary virtual node is created for each target book, the virtual node is stored in memory in a key-value pair data structure, the aggregated metadata is encapsulated into a temporarily existing structured node, and the virtual node is not written into a physical storage device, the virtual node is generated for each target book and includes three types of attributes, including basic attributes, semantic attributes and derived attributes, the basic attributes include ISBN, book name, author, publisher and publication time, the basic attributes are used as the core identifier of the node, the semantic attributes include subject words, abstract keywords and discipline classification, and are used to support semantic association, and the derived attributes include user ratings, price fluctuation values and associated data recommendation degrees, to reflect the dynamic characteristics and associated potential of the book;

[0011] An associated edge between the virtual nodes is automatically constructed through semantic vector calculation to establish a dynamic link, the dynamic link is realized based on two types of operations, including entity association and relationship annotation, the entity association quantifies the semantic association strength between the book node and other entity nodes through semantic vector calculation, and the specific calculation formula is as follows:

[0012] ,

[0013] Wherein, represents a cosine similarity, used to determine whether the entity association is established, represents a semantic vector of the book node, represents a semantic vector of another entity node, represents the i-th component of the vector represents the i-th component of the vector represents the i-th component of the vector represents the i-th component of the vector represents the dot product of two vectors, the book node and the other entity node are automatically linked based on the semantic association strength, the relationship annotation explicitly annotates the relationship type of the associated edge, and dynamically assigns a relationship weight based on the co-occurrence frequency, and the relationship weight is used to quantify the association strength.

[0014] In a preferred embodiment, in S102, the virtual node is used as a temporary computing entity to support the processing and result returning of the word query, and the node is automatically released after the query response is completed. The trigger condition for the automatic release of the node is that, after the user obtains the search and recommendation results, the automatic release process is started after a delay of five seconds, all attribute information of the virtual node in the memory is cleared, and no physical storage is performed. For the secondary query scenario, the repeated collection of metadata is avoided through the cache mechanism. In the first query, the metadata relied on by the virtual node is temporarily stored in the memory cache area. When the secondary query is triggered, the metadata in the cache is directly called to regenerate the virtual node, and there is no need to initiate a collection request to the multi-source data source again.

[0015] When the virtual node attributes are classified and labeled according to stability, the virtual node attributes are divided into two categories according to stability, including static attributes and dynamic attributes, and the time effectiveness of the attributes is labeled. The static attribute refers to the metadata that remains unchanged for a long time, including the author, ISBN, publisher and publication time, which is marked as long-term valid and is generated only in the first metadata collection. The dynamic attribute refers to the metadata that may change in a short time, including the price, user rating and inventory status, which is marked as short-term valid. The time effectiveness label is embedded in the attribute field of the virtual node in the form of a key-value pair.

[0016] When the remaining validity period of the dynamic attribute is insufficient, a new round of metadata collection request is automatically initiated to the corresponding data source to obtain the latest attribute value. After the collection is completed, if there is a difference between the newly obtained attribute value and the original attribute value, the corresponding attribute of the virtual node is immediately updated, and the time effectiveness is re-labeled. If there is no significant difference, the original attribute value is maintained but the time effectiveness label is refreshed.

[0017] In a preferred embodiment, in S103, the user query request is parsed into entity, attribute and relationship triples. The triples are formed through a semantic parsing model, and synonym and near-synonym expansion is realized based on the simple knowledge organization system thesaurus. The expansion logic is that when the subject word of the book node matches a concept in the simple knowledge organization system thesaurus, all synonymous and near-synonymous concepts corresponding to the subject word of the concept are automatically associated. Through the citation data of the academic database, a cited-citing relationship chain is constructed. The basis of the relationship chain is the citation record of the academic database, which contains the cited information, the citation position and the citation purpose of the book. The association logic is to build the association edge of the cited information for each book citation record, and label the citation strength.

[0018] In a preferred embodiment, in S104, the triples generated by entity relationship recognition are mapped to the node attributes of the virtual knowledge graph, the semantic association between the query intention and the graph structure is established, the mapping process needs to ensure semantic consistency, the field is accurately matched through a pre-set attribute mapping table, a sub-graph is dynamically constructed, which contains the target book and the directly associated nodes, the generation range of the sub-graph is defined by the direct association relationship of the query entity, for a single book entity, the sub-graph contains the book node and the directly associated entity nodes, for a category entity, the sub-graph contains all book nodes under the category and the high-frequency entity nodes associated with them;

[0019] All book nodes that meet the triple constraint conditions are filtered out by traversing the virtual sub-graph through the breadth-first search algorithm, the execution logic of the breadth-first search algorithm is: taking the core entity node of the sub-graph as the starting point, traversing the adjacent nodes layer by layer, checking the attributes of each node, and judging whether the attributes meet the attribute constraints of the triples, and if they meet the constraints, marking them as matching nodes;

[0020] The matching nodes are arranged in descending order based on the association strength, the association degree between the nodes and the query is quantified to ensure that the most relevant books are presented first, and the specific calculation formula of the association strength is as follows:

[0021] ,

[0022] Wherein, S represents the association strength, M represents the attribute matching degree, W represents the relationship weight, T represents the dynamic attribute time effectiveness, M represents the attribute matching degree weight, W represents the weight of the relationship weight, T represents the weight of the dynamic attribute time effectiveness, based on the associated path of the virtual graph, personalized recommendation is realized, by mining the multi-hop associated path in the virtual graph, candidate books that have indirect association with the target book are found, multi-hop traversal is performed based on the associated edges of the virtual graph, typical paths include author association path, theme association path and citation association path, through multi-path cross verification, candidate books that have stable association with the target book are screened out, and a candidate pool is provided for personalized recommendation.

[0023] The beneficial effects of the application are: the application realizes low-delay collection through API interface, dynamic attributes are updated in real time according to short-term effective rules, virtual nodes adopt a fast automatic release and reasonable cache period reuse mechanism, which significantly reduces memory occupation, secondary query reuses cache metadata, response speed is obviously improved, server load is effectively reduced, resource redundancy is avoided, deep relationships are mined through cross-source linking and path reasoning, and the accuracy of recommendation is improved. BRIEF DESCRIPTION OF DRAWINGS

[0024] Figure 1 The method flowchart of the application. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the present application will be described clearly and completely in the present application with reference to the accompanying drawings. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.

[0026] In the description of the present application, the terms "first", "second" are used only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0027] In the description of the present application, the term "for example" is used to indicate "as an example, illustration or description". Any embodiment described as "for example" in the present application is not necessarily interpreted as more preferred or more advantageous than other embodiments. The following description is given in order to enable any person skilled in the art to implement and use the present application. In the following description, details are listed for the purpose of explanation. It should be understood that those skilled in the art can realize the present application without using these specific details. In other examples, well-known structures and processes will not be described in detail to avoid unnecessary details making the description of the present application obscure. Therefore, the present application is not intended to be limited to the shown embodiments, but is consistent with the broadest scope in accordance with the principles and characteristics disclosed in the present application.

[0028] As Figure 1 The present embodiment provides a metadata-driven book search and recommendation method based on a virtual knowledge graph, including the following steps:

[0029] S101, receiving a query request, triggering multi-source metadata aggregation, generating a temporary virtual node, and establishing an associated edge between nodes and marking the relationship weight through semantic vector calculation;

[0030] Furthermore, the system receives user-input query requests, limiting them to ISBN, book title, and keywords. The parsing unit employs a differentiated processing mechanism to parse the query requests. For ISBN types, it matches their standard format using regular expressions; for book title types, it uses a string fuzzy matching algorithm to handle input errors; and for keyword types, it extracts core search identifiers using topic term extraction technology. All parsing results are converted into standardized search identifiers. Based on these standardized search identifiers, a distributed metadata collector is triggered to perform real-time aggregation. The data sources are divided into three categories: basic data sources, dynamic data sources, and related data sources. The metadata collector connects to each data source through a standardized API interface to ensure data format uniformity.

[0031] A temporary virtual node is created for each target book. The virtual node is stored in memory as a key-value pair data structure. The aggregated metadata is encapsulated into a temporary structured node and is not written to the physical storage device. The virtual node is generated separately for each target book and includes three types of attributes: basic attributes, semantic attributes, and derived attributes. The basic attributes cover ISBN, book title, author, publisher, and publication time. The basic attributes are used as the core identifier of the node. The semantic attributes include subject terms, abstract keywords, and subject classification to support semantic association. The derived attributes include user rating, price fluctuation value, and related data recommendation degree to reflect the dynamic characteristics and association potential of the book.

[0032] Dynamic links are established by automatically constructing relational edges between virtual nodes through semantic vector computation. These dynamic links are implemented based on two types of operations: entity association and relation annotation. Entity association quantifies the semantic association strength between the book node and other entity nodes through semantic vector computation. The specific calculation formula is as follows:

[0033] ,

[0034] in, Cosine similarity is used to determine whether an entity association exists. The semantic vector representing the book node. The semantic vector representing other entity nodes, Representing vectors The i-th component, Representing vectors The i-th component, where k represents the vector dimension. It represents the dot product of two vectors, automatically links book nodes with other entity nodes based on semantic association strength, explicitly marks the relationship type on the associated edges, and dynamically assigns relationship weights based on co-occurrence frequency, and quantifies the association strength based on the relationship weights.

[0035] It should be noted that the basic data source includes the ISBN, author, publication time and other authoritative identification information provided by the library OPAC system, the content abstract and the directory provided by the publisher database, and the accuracy of the basic attributes of the book is ensured. The dynamic data source includes the current price and inventory status returned by the e-commerce platform in real time, the user rating and tag cloud provided by the Du community, reflecting the real-time market feedback of the book, the citation relationship provided by the academic database through the association data source, the recent discussion heat provided by the social platform, and the click and purchase associated data provided by the user behavior log, supporting the correlation analysis of subsequent recommendations. Based on the establishment of dynamic link, the dispersed virtual nodes form a meshed association structure, which not only retains the semantic logic between entities, but also adjusts the change of the association relationship through weight dynamic adjustment (such as the association between new books and classic books increases with the increase of user behavior).

[0036] S102, the virtual node is automatically released after the query response, the cache metadata is reused for secondary query, and the virtual node attributes are classified and labeled according to stability;

[0037] Further, the virtual node is used as a temporary computing entity to support the processing and result returning of word query. After the query response is completed, the node is automatically released. The trigger condition for automatic release of the node is that after the user obtains the search and recommendation results, a five-second delay is started to initiate the automatic release process, and all attribute information (including basic attributes, semantic attributes, and derived attributes) of the virtual node in the memory is cleared without any physical storage. For secondary query scenarios (such as user page turning, filtering, and sorting operations), the cache mechanism is used to avoid repeated collection of metadata. In the first query, the metadata relied on by the virtual node is temporarily stored in the memory cache area. When the secondary query is triggered, the metadata in the cache is directly called to regenerate the virtual node, and there is no need to initiate a collection request to the multi-source data source again.

[0038] The virtual node attributes are classified and labeled according to stability, and the virtual node attributes are divided into two categories according to stability, including static attributes and dynamic attributes, and the time efficiency of the attributes is labeled. The static attribute refers to the metadata that remains unchanged for a long time, including the author, ISBN, publisher and publication time, which is marked as long-term valid, and is generated only in the first metadata collection. The dynamic attribute refers to the metadata that may change in a short time, including the price, user rating and inventory status, which is marked as short-term valid. The time efficiency label is embedded in the attribute field of the virtual node in the form of key-value pair.

[0039] When the remaining validity period of the dynamic attribute is insufficient (e.g., less than ten seconds), a new round of metadata collection request is automatically initiated to the corresponding data source to obtain the latest attribute value (e.g., the current real-time price, the updated user score), after the collection is completed, if the newly obtained attribute value is different from the original attribute value, the corresponding attribute of the virtual node is immediately updated, and the time limit is re-labeled, if there is no significant difference, the original attribute value is maintained but the time limit label is refreshed.

[0040] It should be noted that the cooperation of the time limit label and the node release mechanism forms a closed-loop management of the whole life cycle of the virtual node, which not only guarantees the data validity, but also realizes the optimal allocation of resources. The long-term validity of the static attribute is adapted to the node release mechanism. Since the static attribute does not need to be updated frequently, its cache validity period (thirty minutes) covers the secondary query window period after the node is released, ensuring that accurate static information can still be reused when the secondary node is generated. The short-term validity of the dynamic attribute is linked with the node release mechanism through the update mechanism. If the virtual node is released within the validity period of the dynamic attribute, the dynamic metadata in the cache will be marked as to be updated synchronously with the node release. When the secondary query triggers the node reconstruction, the latest dynamic attribute will be preferentially collected, avoiding the use of expired cache.

[0041] S103, parse the query into triples, build a citation chain with the simple knowledge organization system thesaurus and citation data, and realize accurate association across source entities;

[0042] Further, the user query request is parsed into entity, attribute and relationship triples, the triples are constructed through a semantic analysis model, synonym and near-synonym expansion is realized based on the simple knowledge organization system thesaurus, and the expansion logic is: when the subject word of a book node matches a certain concept in the simple knowledge organization system thesaurus, automatically associate all synonymous and near-synonymous concepts corresponding to the subject word of the concept, build a cited-cited relationship chain through the citation data of the academic database, the relationship chain is based on the citation record of the academic database, the citation record contains the cited information, the citation position and the citation purpose of the book, and the association logic is to build an associated edge for each book citation record, and label the citation strength.

[0043] It should be noted that the composition of triples follows strict semantic logic, entity refers to the core object involved in the query, attribute refers to the characteristic dimension of the entity, and relationship refers to the association between the entity and the attribute. The relationship chain reveals the knowledge association between books, enabling the recommendation of books related to the query book through the relationship chain, and providing a complete resource context from the foundation to the application for academic research. The relationship chain is built from three dimensions of author organization, theme semantics and knowledge inheritance, and cooperates with the triples of entity relationship recognition to form a multi-dimensional virtual graph association network.

[0044] S104, the search engine returns the matching results through semantic analysis, sub-spectrum generation and correlation strength, and the recommendation engine generates recommendations based on multi-hop path reasoning;

[0045] Further, the entity relationship recognition generates triples mapped to the node attributes of the virtual knowledge graph, establishes the semantic association between the query intent and the graph structure, and ensures semantic consistency through the mapping process. Precise field matching is achieved through a pre-set attribute mapping table, a sub-spectrum containing target books and directly associated nodes is dynamically constructed, and the generation range of the sub-spectrum is defined by the direct association relationship of the query entity. For a single book entity, the sub-spectrum contains the book node and its directly associated entity nodes. For a category entity, the sub-spectrum contains all book nodes under the category and their commonly associated high-frequency entity nodes.

[0046] All book nodes that meet the triple constraint conditions are filtered out by traversing the virtual sub-spectrum through the breadth-first search algorithm. The execution logic of the breadth-first search algorithm is as follows: taking the core entity node of the sub-spectrum as the starting point, traversing the adjacent nodes layer by layer (the first layer is the directly associated node, and the second layer is the associated node of the associated node), checking the attributes of each node, and determining whether it meets the attribute constraints of the triple. If it meets the constraints, it is marked as a matching node.

[0047] The matching nodes are arranged in descending order based on the correlation strength, the correlation degree between the nodes and the query is quantified to ensure that the most relevant books are presented first, and the specific calculation formula of the correlation strength is as follows:

[0048] ,

[0049] Where S represents the correlation strength, M represents the attribute matching degree, W represents the relationship weight, T represents the dynamic attribute time effectiveness, M represents the attribute matching degree weight, W represents the relationship weight weight, T represents the dynamic attribute time effectiveness weight, based on the associated path of the virtual graph, individualized recommendations are realized, by mining the multi-hop associated path in the virtual graph, candidate books that have indirect associations with target books are discovered, multi-hop traversal is performed based on the associated edges of the virtual graph, typical paths include author association path, theme association path and citation association path, through multi-path cross verification, candidate books that have stable associations with target books are selected, providing a candidate pool for individualized recommendations.

[0050] It should be noted that the matching process needs to consider both integrity and efficiency: the breadth-first search algorithm has a traversal depth limit of 2 layers (to avoid excessive expansion leading to redundant calculations), the adjacent nodes of the matched nodes are preferentially checked (such as the same author books of a certain computer textbook), repeated calculations are reduced through pruning strategies (such as skipping nodes that have been checked and do not meet the conditions), and the output of the subgraph matching is the set of all nodes that meet the constraint conditions, which is the direct input for result sorting.

[0051] It should be noted that the author association path is: book A -> same author -> other works -> book B, the theme association path is: book C -> high similarity theme -> related theme book -> book D, and the citation association path is: book E -> frequently cited -> cited book -> book F.

[0052] It should be noted that in the above embodiments, the description of each embodiment has its own focus, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0053] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) containing computer usable program code.

[0054] The present application is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions, and the combination of flows and / or blocks in the flowcharts and / or block diagrams. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing devices to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices produce a machine that implements the flow Figure 1 The function specified in one flow or multiple flows and / or blocks. Figure 1 The function specified in one flow or multiple flows and / or blocks.

[0055] These computer program instructions can also be stored in a computer readable memory that can guide the computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer readable memory produce a product including instruction devices, which implement the flow Figure 1 The function specified in one flow or multiple flows and / or blocks. Figure 1 The function specified in one flow or multiple flows and / or blocks.

[0056] These computer program instructions can also be loaded into computer or other programmable data processing devices, so that a series of operations steps are performed on the computer or other programmable devices to generate computer-implemented processes, thus the instructions executed on the computer or other programmable devices provide the function of realizing the processes specified in the flowcharts Figure 1 one or more flows and / or blocks Figure 1 one or more blocks or steps of the functions specified in the flowcharts or block diagrams.

[0057] Although the preferred embodiments of the application have been described, those skilled in the art will be able to make additional modifications and variations to these embodiments without departing from the spirit and scope of the application. Accordingly, it is intended that the appended claims be construed to include all such modifications and variations as fall within the scope of the application.

[0058] Obviously, various modifications and changes are possible in the present application without departing from the spirit and scope of the application. Accordingly, it is intended that the present application embrace all modifications and changes that fall within the scope of the claims and their equivalents.

Claims

1. A metadata driven book search and recommendation method based on virtual knowledge graph, characterized in that, The method comprises the following steps: S101, receiving a query request, triggering multi-source metadata aggregation, generating a temporary virtual node, and establishing an associated edge between nodes and marking the relationship weight through semantic vector calculation; A temporary virtual node is created for each target book. The virtual node is stored in memory as a key-value data structure. The aggregated metadata is encapsulated into a temporarily existing structured node and is not written into a physical storage device. The virtual node is generated for each target book and includes three types of attributes, namely basic attributes, semantic attributes, and derived attributes. The basic attributes include ISBN, book name, author, publisher, and publication time. The basic attributes are used as the core identifier of the node. The semantic attributes include subject words, abstract keywords, and discipline classification, which are used to support semantic association. The derived attributes include user ratings, price fluctuation values, and associated data recommendation degrees. S102, the virtual node is automatically released after the query response, and the cached metadata is reused for secondary queries. The virtual node attributes are classified and labeled according to their stability. The virtual node attributes are divided into two categories according to their stability, including static attributes and dynamic attributes. The time limit is marked for each attribute. Static attributes refer to metadata that remains unchanged for a long time, including author, ISBN, publisher, and publication time. They are marked as long-term valid and only generated during the first metadata collection. Dynamic attributes refer to metadata that may change in a short period of time, including price, user ratings, and inventory status. They are marked as short-term valid. The time limit is embedded in the attribute field of the virtual node in the form of a key-value pair. S103, parse the query into triples, build the citation chain with simple knowledge organization system thesaurus and citation data, and realize accurate association across source entities. S104, the search engine returns the matching results based on semantic analysis, sub-graph generation, and correlation strength. The recommendation engine generates recommendations based on multi-hop path reasoning. 2.The virtual knowledge graph based metadata driven book search and recommendation method according to claim 1, characterized in that, In S101, the user input query request is received, and the query request is limited to ISBN, book name, and keywords. The parsing unit uses a differentiated processing mechanism to parse the query request. For ISBN type, the standard format is matched through regular expression. For book name type, a string fuzzy matching algorithm is used to handle input errors. For keyword type, the core retrieval identifier is extracted through topic word extraction technology. All parsed results are converted into standardized retrieval identifiers. Based on the standardized retrieval identifier, the distributed metadata collector is triggered to perform real-time aggregation. The data sources are divided into three categories, including basic data sources, dynamic data sources, and associated data sources. The metadata collector connects with each data source through a standardized API interface to ensure data format uniformity. 3.The virtual knowledge graph based metadata driven book search and recommendation method according to claim 2, characterized in that, The associated edges between virtual nodes are automatically constructed through semantic vector calculation to establish dynamic links. Dynamic links are implemented based on two types of operations, including entity association and relationship labeling. Entity association quantifies the semantic association strength between book nodes and other entity nodes through semantic vector calculation. The specific calculation formula is as follows: ， wherein, denotes cosine similarity, used to determine whether the entity association is valid, denotes the semantic vector of the book node, denotes the semantic vector of the other entity node, denotes the i-th component of the vector denotes the i-th component of the vector denotes the i-th component of the vector denotes the i-th component of the vector, k denotes the vector dimension, denotes the dot product of two vectors, automatically links the book node with the other entity node based on the semantic association strength, explicitly marks the relationship type for the association edge by relationship marking, dynamically assigns the relationship weight based on the co-occurrence frequency, and quantifies the association strength based on the relationship weight. 4.The virtual knowledge graph based metadata driven book search and recommendation method according to claim 1, wherein, In S102, the virtual node is used as a temporary computing entity to support the processing and result returning of the word query, and the node is automatically released after the query response is completed. The trigger condition for the automatic release of the node is that, after the user obtains the search and recommendation results, the automatic release process is started after a delay of five seconds, all attribute information of the virtual node in the memory is cleared, and no physical storage is performed. For the secondary query scenario, the repeated collection of metadata is avoided through the cache mechanism. When the first query is performed, the metadata relied on by the virtual node is temporarily stored in the memory cache area. When the secondary query is triggered, the metadata in the cache is directly called to regenerate the virtual node, and there is no need to initiate a collection request to the multi-source data source again. 5.The virtual knowledge graph based metadata driven book search and recommendation method according to claim 4, characterized in that, When the remaining valid period of the dynamic attribute is insufficient, a new round of metadata collection request is automatically initiated to the corresponding data source to obtain the latest attribute value. After the collection is completed, if there is a difference between the newly obtained attribute value and the original attribute value, the corresponding attribute of the virtual node is immediately updated, and the time limit is re-labeled. If there is no significant difference, the original attribute value is maintained but the time limit is refreshed. 6.The virtual knowledge graph based metadata driven book search and recommendation method according to claim 1, wherein, In S103, the user query request is parsed into entity, attribute, and relationship triplets. The triplets are constructed through a semantic analysis model, and synonym and near-synonym expansion is realized based on a simple knowledge organization system thesaurus. The expansion logic is that, when the subject word of a book node matches a certain concept in the simple knowledge organization system thesaurus, all synonymous and near-synonymous concepts corresponding to the subject word of the concept are automatically associated. Through the citation data of the academic database, a cited-cited relationship chain is constructed. The relationship chain is based on the citation record of the academic database, and the citation record includes the cited information, the citation position, and the citation purpose of the book. The association logic is that, for each book citation record, a cited association edge is constructed, and the citation strength is labeled. 7.The virtual knowledge graph based metadata driven book search and recommendation method according to claim 1, wherein, In S104, the triplets generated by entity relationship identification are mapped to the node attributes of the virtual knowledge graph to establish the semantic association between the query intent and the graph structure. The mapping process needs to ensure semantic consistency, and accurate field matching is realized through a pre-set attribute mapping table. A sub-graph is dynamically constructed, which contains the target book and the directly associated nodes. The generation range of the sub-graph is defined by the direct association relationship of the query entity. For a single book entity, the sub-graph contains the book node and the directly associated entity nodes. For a category entity, the sub-graph contains all book nodes under the category and the high-frequency entity nodes commonly associated with them. 8.The virtual knowledge graph based metadata driven book search and recommendation method according to claim 7, characterized in that, All book nodes that meet the triplet constraint conditions are selected by traversing the virtual sub-graph through a breadth-first search algorithm. The execution logic of the breadth-first search algorithm is that the core entity node of the sub-graph is taken as the starting point, adjacent nodes are traversed layer by layer, the attributes of each node are verified, and it is determined whether the attributes meet the attribute constraints of the triplets. If they meet the constraints, the nodes are marked as matching nodes. 9.The virtual knowledge graph based metadata driven book search and recommendation method according to claim 7, characterized in that, The matching nodes are arranged in descending order based on the association strength, the association degree between the nodes and the query is quantified, and the most relevant books are ensured to be presented first. The specific calculation formula of the association strength is as follows: ， Wherein, S represents the correlation strength, M represents the attribute matching degree, W represents the relationship weight, T represents the dynamic attribute time effectiveness, alpha represents the attribute matching degree weight, beta represents the relationship weight weight, gamma represents the dynamic attribute time effectiveness weight, the correlation path based on the virtual graph is used for realizing personalized recommendation, candidate books which have indirect correlation with the target book are found by mining the multi-hop correlation path in the virtual graph, multi-hop traversal is carried out based on the correlation edge of the virtual graph, the typical path includes the author correlation path, the theme correlation path and the citation correlation path, the candidate books which have stable correlation with the target book are screened out through multi-path cross verification, and the candidate pool is provided for personalized recommendation.

Images