Library intelligent searching and sorting method and system

Abstract

The invention relates to the technical field of information retrieval, and provides an intelligent searching and sorting method and system for a library. Obtaining environment parameters of a search terminal and query statements and historical retrieval records input by a user; generating a dynamic environment matrix based on the environment parameters, and generating an intention vector based on a query statement and the historical retrieval record; constructing a knowledge graph based on the metadata of the library, extracting a literature feature vector from the knowledge graph, and performing linear processing on the intention vector and the literature feature vector to generate a literature association graph; generating a comprehensive weight vector based on the statistical parameters extracted from the literature association map and the dynamic environment matrix; and performing comprehensive evaluation based on the comprehensive weight vector and the score of the literature in the preset dimension, and displaying a search result on a user interface according to the retrieval score of the literature. According to the scheme, the search result is highly matched with the user requirement and the scene, and the accuracy and the search efficiency of library intelligent search are remarkably improved.

Description

Technical Field

[0001] This application relates to the field of information retrieval technology, and more specifically, to a library intelligent search and sorting method and system. Background Technology

[0002] With the exponential growth of digital resources such as e-books and academic papers, users' demands for personalized search results are becoming increasingly significant. For example, medical researchers may be more interested in the latest clinical trial literature, while students need classic textbooks. Meanwhile, traditional library keyword-based search methods are no longer sufficient to meet user needs. For instance, a university library's digital resources have seen an annual growth rate of 30%. This surge in data volume has caused search logic and methods to lag behind the ever-increasing resource volume and user demands, extending the average search time from 5 minutes to 15 minutes. There is an urgent need for intelligent search and ranking to address the current problems of low efficiency and accuracy in library searches. Summary of the Invention

[0003] This application provides a library intelligent search and sorting method and system, which can at least partially solve the problem of low retrieval efficiency and accuracy in libraries.

[0004] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part from practice of this application.

[0005] According to one aspect of this application, a library intelligent search ranking method is provided, comprising: acquiring environmental parameters of a search terminal, a query statement input by a user, and historical search records; generating a dynamic environment matrix based on the environmental parameters, and generating an intent vector based on the query statement and the historical search records; constructing a knowledge graph based on library metadata, and extracting document feature vectors from the knowledge graph, and generating a document association graph by linearly processing the intent vector and the document feature vectors; generating a comprehensive weight vector based on statistical parameters extracted from the document association graph and the dynamic environment matrix; performing a comprehensive evaluation based on the comprehensive weight vector and the document's score in a preset dimension to determine the document's search score, and displaying search results on a user interface according to the document's search score.

[0006] In this application, based on the aforementioned scheme, obtaining the environmental parameters of the search terminal, the query statement input by the user, and historical search records includes: retrieving the environmental parameters of the search terminal based on the identifier of the search terminal; obtaining the query statement input by the user when a query command triggered by the user is detected; and retrieving the user's historical search records from the database based on the user's account identifier.

[0007] In this application, based on the aforementioned scheme, generating an intent vector based on the query statement and the historical retrieval records includes: performing word segmentation on the query statement to generate query words and assembling a first sequence; performing vectorization on the historical retrieval records to generate a second sequence corresponding to the historical behavior; capturing the contextual relationships in the first sequence through an encoder to generate word vectors of a preset dimension; extracting flow information in the second sequence through a gating model to generate a hidden state vector; and combining the word vectors and the hidden state vectors to generate the intent vector.

[0008] In this application, based on the aforementioned scheme, the step of constructing a knowledge graph based on library metadata and extracting document feature vectors from the knowledge graph, and generating a document association graph by linearly processing the intent vector and the document feature vectors, includes: constructing a knowledge graph based on library metadata and extracting document feature vectors from the knowledge graph; generating intent neighborhoods based on the intent vectors; determining a document set consisting of intent files related to user intents based on the relationship parameters between the intent vectors, the intent neighborhoods, and the document feature vectors; and constructing a document association graph based on the intent nodes formed by the intent vectors and the document nodes in the document set.

[0009] In this application, based on the aforementioned scheme, the step of generating a comprehensive weight vector based on the statistical parameters extracted from the document association graph and the dynamic environment matrix includes: extracting statistical parameters of a preset dimension from the document association graph; determining the dimension importance coefficient corresponding to the preset dimension based on the statistical parameters; determining the dimension factor corresponding to the preset dimension based on the statistical parameters, the dimension importance coefficient, and the dynamic environment matrix; and combining the dimension factors corresponding to the preset dimension to generate a comprehensive weight vector.

[0010] In this application, based on the aforementioned scheme, the step of comprehensively evaluating the document based on the comprehensive weight vector and the document's score in a preset dimension to determine the document's retrieval score includes: determining a first score for the document in each preset dimension based on the comprehensive weight vector and the document's score in the preset dimension; calculating a second score based on the first score using a preset nonlinear enhancement coefficient; and determining the document's retrieval score based on the linear sum between the first score and the second score.

[0011] In this application, based on the aforementioned scheme, determining the first score of the document in each preset dimension based on the comprehensive weight vector and the document's score in the preset dimensions includes: determining the first score of the document in each preset dimension based on the comprehensive weight vector and the document's score in the preset dimensions using the following formula. for:

[0012] in, Indicates the preset dimension i The corresponding dimension factor, Documents j In the i Wei's rating, i and d These represent the identifier and total number of the preset dimensions, respectively.

[0013] In this application, based on the aforementioned scheme, the step of calculating the first score based on a preset nonlinear enhancement coefficient to determine the second score includes: calculating the first score based on the preset nonlinear enhancement coefficient, and determining the second score using the following formula. for:

[0014] in, Indicates the preset dimension k The corresponding dimension factor, Documents j In the k Wei's rating, k and d These represent the total number of identifiers for the preset dimensions.

[0015] In this application, based on the aforementioned scheme, determining the retrieval score of the document according to the linear sum between the first score and the second score includes: determining the document score using the following formula based on the linear sum between the first score and the second score. j Search rating for:

[0016] in, This represents the nonlinear enhancement coefficient determined experimentally.

[0017] In this application, based on the aforementioned scheme, the step of displaying search results on the user interface according to the search score of the document includes: sorting the documents according to the search score of the document to generate a sorted list; and displaying search results on the user interface based on the sorted list.

[0018] According to one aspect of this application, a library intelligent search and sorting system is provided, comprising: The acquisition unit is used to acquire the environmental parameters of the search terminal, as well as the query statement and historical search records entered by the user; The intent unit is used to generate a dynamic environment matrix based on the environment parameters and to generate an intent vector based on the query statement and the historical search records. The document unit is used to construct a knowledge graph based on the library's metadata, extract document feature vectors from the knowledge graph, and generate a document association graph by linearly processing the intent vector and the document feature vectors. The weighting unit is used to generate a comprehensive weight vector based on the statistical parameters extracted from the document association map and the dynamic environment matrix. The display unit is used to perform a comprehensive evaluation based on the comprehensive weight vector and the scores of the document in a preset dimension, determine the retrieval score of the document, and display the search results on the user interface according to the retrieval score of the document.

[0019] According to one aspect of this application, a computer-readable medium is provided having a computer program stored thereon, which, when executed by a processor, implements the library intelligent search and sorting method as described in the above embodiments.

[0020] According to one aspect of this application, an electronic device is provided, comprising: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, cause the one or more processors to implement the library intelligent search and sorting method as described in the above embodiments.

[0021] According to one aspect of this application, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the library intelligent search and sorting method provided in the various alternative implementations described above.

[0022] In this application's technical solution, environmental parameters of the search terminal, user-input query statements, and historical search records are obtained. A dynamic environment matrix is ​​generated based on the environmental parameters, and an intent vector is generated based on the query statements and historical search records. A knowledge graph is constructed based on library metadata, and document feature vectors are extracted from the knowledge graph. A document association graph is generated by linearly processing the intent vector and the document feature vectors. A comprehensive weight vector is generated based on statistical parameters extracted from the document association graph and the dynamic environment matrix. A comprehensive evaluation is performed based on the comprehensive weight vector and the document's score in a preset dimension to determine the document's search score. Based on the document's search score, the search results are displayed on the user interface. This application's technical solution, through the synergy of environmental awareness, semantic modeling, dynamic weighting, and nonlinear processing, achieves a high degree of adaptation between search results and user needs and scenarios, significantly improving the accuracy and efficiency of intelligent library retrieval.

[0023] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0024] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0025] Figure 1 The flowchart of a library intelligent search sorting method in one embodiment of this application is illustrated schematically.

[0026] Figure 2 The flowchart illustrating the generation of an intent vector is shown in one embodiment of this application.

[0027] Figure 3 The illustration shows a schematic diagram of a library intelligent search and sorting system in one embodiment of this application.

[0028] Figure 4 A schematic diagram of the structure of a computer system suitable for implementing an electronic device according to an embodiment of the present application is shown. Detailed Implementation

[0029] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided to make this application more comprehensive and complete, and to fully convey the concept of the exemplary embodiments to those skilled in the art.

[0030] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a thorough understanding of embodiments of this application. However, those skilled in the art will recognize that the technical solutions of this application can be practiced without one or more of the specific details, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known methods, apparatuses, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this application.

[0031] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0032] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, while others may be combined or partially combined. Therefore, the actual execution order may vary depending on the actual situation.

[0033] The implementation details of the technical solution of this application are described below: Figure 1 A flowchart of a library intelligent search and sorting method according to an embodiment of this application is shown. (Refer to...) Figure 1 As shown, the library's intelligent search and sorting method includes at least steps S110 to S150, which are detailed below: S110, obtain the environmental parameters of the search terminal, as well as the query statement and historical search records entered by the user.

[0034] When acquiring environmental parameters of the search terminal, real-time data is collected through the terminal device interface or browser environment variables. For example, information such as the operating system and terminal identification code can be obtained through Python platform modules. After standardization, a dynamic environment matrix including dimensions such as illumination and noise is constructed. At the same time, user query statements are transformed into structured input through word segmentation, part-of-speech tagging, and semantic parsing. Combined with the storage and retrieval mechanism of historical search records, such as the paper platform managing search history through login accounts and supporting classification, sorting, and export operations, an intent vector is finally generated through a weighted fusion strategy. The environment matrix adjusts the weight coefficients in real time to correct the intent vector, ensuring that the search intent and scenario requirements are accurately matched, forming a multimodal feature representation that takes into account both explicit user needs and implicit environmental constraints.

[0035] In one embodiment of this application, obtaining the environmental parameters of the search terminal, the user-input query statement, and historical search records includes: Based on the identifier of the search terminal, retrieve the environmental parameters of the search terminal; When a query command triggered by the user is detected, the query statement entered by the user is obtained; Based on the user's account identifier, retrieve the user's historical search records from the database.

[0036] In one embodiment of the application, the environmental parameters of the search terminal are retrieved based on the identifier of the search terminal. Specifically, the environmental parameters include information such as device type (mobile phone, computer or tablet), operating system (iOS, Android) and network status (Wi-Fi, 5G).

[0037] In one embodiment of the application, upon detecting a query command triggered by a user, the query statement entered by the user is obtained. For example, recommending a science fiction novel to be published in 2025.

[0038] In one embodiment of the application, the user's historical search records are retrieved from the database based on the user's account identifier. For example, keywords searched by the user in the past 30 days, such as Nobel Prize literature or AI-themed novels.

[0039] S120, Generate a dynamic environment matrix based on the environment parameters, and generate an intent vector based on the query statement and the historical retrieval records.

[0040] In this embodiment, when generating a dynamic environment matrix based on environmental parameters, the operating parameters of the retrieval device are first obtained through the device identifier of the retrieval device. A dynamic environment matrix is ​​then generated based on a preset matrix model, which dynamically reflects the operating status of the retrieval device used by the user. Simultaneously, combining the semantic parsing results of the query statement, keyword vectors are extracted through a pre-trained model and combined with statistical features of historical search records, such as high-frequency search terms and document type preferences. A weighted fusion strategy is then used to generate an intent vector. The environment matrix adjusts the weight coefficients to continuously correct the intent vector, ensuring an accurate match between the search intent and the scenario requirements. Ultimately, this results in a multimodal feature representation that considers both explicit user needs and implicit environmental constraints.

[0041] like Figure 2 As shown, in one embodiment of this application, an intent vector is generated based on the query statement and the historical retrieval records, including: S210, the query statement is segmented into words to generate query words and form a first sequence; S220, the historical retrieval records are vectorized to generate a second sequence corresponding to the historical behavior; S230, The encoder captures the contextual relationships in the first sequence and generates word vectors of a preset dimension; S240, extract the flow information in the second sequence through the gating model to generate a hidden state vector; S250, combine the word vector and the hidden state vector to generate the intent vector.

[0042] In one embodiment of this application, the query statement is segmented into words to generate query segments and form a first sequence. In this embodiment, a Chinese word segmentation tool can be used to segment the query statement to obtain a query segmentation sequence. ,in, These represent query word segmentation, k The identifier indicating the query segmentation. m This indicates the total number of query words.

[0043] In one embodiment of this application, the historical retrieval records are vectorized to generate a second sequence corresponding to the historical behavior. ,in, These represent historical search word segments. T Indicates a time window.

[0044] In one embodiment of this application, an encoder captures the contextual relationships in the first sequence to generate word vectors of a preset dimension. Specifically, each word segmentation... Input the pre-trained word vector encoder BET to generate word vectors of a preset dimension. By capturing the contextual relationships between word segments through a bidirectional encoder, the semantic loss problem of traditional unidirectional encoding is solved.

[0045] In one embodiment of this application, flow information in the second sequence is extracted using a gating model to generate a hidden state vector. Specifically, the second sequence of historical behavior is... h The input-gated model TSM controls the flow of information through forget gates, input gates, and output gates, preserving long-term dependencies in historical retrieval records and generating a sequence of hidden states. ,in, These represent the hidden state components.

[0046] In one embodiment of this application, the word vectors and the hidden state vectors are combined to generate an intent vector. I for:

[0047] in, This indicates maximization processing. k The identifier indicating the query segmentation. m This indicates the total number of query words. This represents the keyword weights calculated using the self-attention mechanism. Indicates query word segmentation; This represents the preset time decay coefficient. v This represents the generated alternative intent vector. T Indicates a time window.

[0048] In the above process, parameters such as terminal device type, operating status, and timestamps are integrated through a dynamic environment matrix, enabling the ranking strategy to adapt to different scenarios, such as fast browsing on mobile devices or in-depth research on computers. By combining semantic parsing of query statements with vector modeling of historical search records, intent vectors can accurately capture users' explicit needs and implicit preferences, such as in high-frequency search areas, avoiding the semantic ambiguity problems of traditional keyword matching.

[0049] S130: Construct a knowledge graph based on the library's metadata, extract document feature vectors from the knowledge graph, and generate a document association graph by linearly processing the intent vector and the document feature vectors.

[0050] When constructing a knowledge graph based on library metadata, the first step is to acquire the library's metadata, including structured information such as document identifiers, object identifiers, titles, authors, keywords, and publication dates. This metadata is then transformed into entities (e.g., document nodes, author nodes) and relationships (e.g., citations, belonging) in the graph, stored in a graph database to form a semantic association network. Subsequently, a pre-trained model encodes the document text content into semantic feature vectors, and combines this with structured data such as time and citation counts to generate multi-dimensional feature vectors. Finally, through linear weighting or matrix operations, such as calculating the cosine similarity between the user intent vector and the document feature vector, a document association graph containing nodes (documents / intents) and weighted edges is constructed. The edge weights reflect the degree of matching between the document and the user's needs, providing a structured semantic basis for subsequent ranking.

[0051] In one embodiment of this application, a knowledge graph is constructed based on library metadata, and document feature vectors are extracted from the knowledge graph. A document association graph is generated by linearly processing the intent vector and the document feature vectors, including: A knowledge graph is constructed based on the library's metadata, and document feature vectors are extracted from the knowledge graph. An intent neighborhood is generated based on the intent vector, and a document set consisting of intent files related to the user's intent is determined based on the relationship parameters between the intent vector, the intent neighborhood, and the file feature vector. Based on the intent nodes formed by the intent vectors and the document nodes in the document set, a document association graph is constructed.

[0052] In one embodiment of this application, a knowledge graph is constructed based on the library's metadata, and document feature vectors are extracted from the knowledge graph. A vector extraction model is used to convert document titles and abstracts into vectors, which serve as document feature vectors. For example, the feature vector for the document "Applications of AI in Medicine" is: =[0.1,0.8,0.5], where 0.1,0.8,0.5 correspond to the feature weights of the technology, medicine, and application themes, respectively.

[0053] In one embodiment of this application, based on the intent vector, through K The nearest neighbor algorithm generates the intent neighborhood corresponding to the intent vector, in order to filter the set of documents most relevant to the user's intent.

[0054] Subsequently, based on the relationship parameters between the intent vector, the intent neighborhood, and the file feature vector... for:

[0055] in, This represents the sharpening factor determined experimentally. I Represents the intent vector. They represent respectively j、k The corresponding feature vector; Indicates the intention neighborhood, i Indicates the intention component, This indicates the preset attenuation rate. t Indicates the time factor. exp This represents the natural exponential function.

[0056] In one embodiment of this application, a set of documents related to user intent is determined by comparing the size of the relation parameter with a set threshold. For example, the intent vector corresponding to the relation parameter being greater than or equal to the set threshold is determined, the intent documents corresponding to the intent vector are determined, and a set of documents consisting of intent files related to user intent is generated.

[0057] In one embodiment of this application, a document association graph is constructed based on the intent nodes formed by the intent vectors and the document nodes in the document set. G =( V,E,W Among them, nodes V Includes intent nodes and document nodes, edges E Connect the intent node and the document node using relation parameters. This represents the weight between the edges formed by two nodes, and is expressed through a weight matrix. W Store the weight values ​​of all edges.

[0058] For example, intent node I =[0.3,0.7], indicating user preference for the intersection of AI and medicine. The document feature vector extracted from the library's metadata-constructed knowledge graph includes document nodes. f 1=[0.2,0.8], Document Node f 2 = [0.1, 0.5], calculated as follows =0.7, =0.3, forming the edge weights of the literature association graph.

[0059] The above process enhances semantic association capabilities through the construction of knowledge graphs and document feature vectors. The knowledge graph, generated from library metadata, links documents to entities such as authors, institutions, and topics, forming a semantic network. Document feature vectors integrate multimodal information such as text content and citation relationships, enabling the document association graph to move beyond simple co-occurrence relationships and uncover deeper academic connections, such as documents from different subfields within the same research direction, thereby improving the relevance of recommendations.

[0060] S140, Based on the statistical parameters extracted from the document association map and the dynamic environment matrix, a comprehensive weight vector is generated.

[0061] In this embodiment, when generating a comprehensive weight vector based on the statistical parameters of the document association graph and the dynamic environment matrix, the mean and variance of each dimension feature (such as semantic similarity and timeliness) are first extracted from the semantic association graph, and the dimension importance coefficient is determined by the statistical average method or the coefficient of variation method. At the same time, combined with real-time parameters such as user dwell time and device type in the dynamic environment matrix, the weight sensitivity is adjusted by using a nonlinear enhancement coefficient, and finally the comprehensive weight is calculated. The sine function introduces periodic excitation to balance the weight distribution of popular and long-tail documents, ensuring that the ranking results reflect both the intrinsic quality of the documents and adapt to the real-time needs of users.

[0062] In one embodiment of this application, a comprehensive weight vector is generated based on the statistical parameters extracted from the document association graph and the dynamic environment matrix, including: Statistical parameters of a preset dimension are extracted from the document association graph, and the dimension importance coefficient corresponding to the preset dimension is determined based on the statistical parameters. Based on the statistical parameters, the dimension importance coefficients, and the dynamic environment matrix, the dimension factors corresponding to the preset dimension are determined. The dimension factors corresponding to the preset dimension are combined to generate a comprehensive weight vector.

[0063] In one embodiment of this application, statistical parameters of a preset dimension are extracted from the document association graph. For example, from the document association graph... G Extract the feature mean of the preset dimension. and variance For example, the average value and fluctuation range of the semantic similarity dimension.

[0064] Then, by statistical averaging, the dimension importance coefficient corresponding to the preset dimension is determined based on the statistical parameters. By adjusting the importance coefficients of the dimensions, we ensure a reasonable allocation of weights among the three dimensions of semantics, timeliness, and popularity, avoiding the dominance of a single dimension, such as sorting solely by popularity, which could lead to new documents being buried.

[0065] In one embodiment of this application, the user's page dwell time and language temperature parameters are extracted from the dynamic environment matrix. Based on the statistical parameters, the dimension importance coefficient, and the page dwell time and language temperature parameters, the preset dimension is determined. i Corresponding dimensional factors for:

[0066] in, They represent the first i The mean and variance corresponding to the preset dimensions. They represent the first j The mean, variance, and dimension importance coefficients corresponding to the preset dimensions. d Indicates the number of preset dimensions; This indicates the duration of a user's page stay, as obtained through front-end tracking. This indicates the preset language temperature parameter.

[0067] In one embodiment of this application, the dimension factors corresponding to the preset dimension are combined to generate a comprehensive weight vector. For example, the comprehensive weight vector generated by combining the dimension factors is... ,For example, (0.4, 0.3, 0.3) represents the dimensional factors of semantic dimension, timeliness dimension, and popularity dimension as 40%, 30%, and 30%, respectively. Through the above process, a dynamic balance of multi-dimensional weights is achieved, providing accurate weight configuration for subsequent ranking. This enables dynamic allocation of dimensions based on information processing and real-time user feedback, ensuring that the results not only meet user intent but also possess real-time performance and diversity.

[0068] The above process achieves multi-objective balance through the dynamic generation of a comprehensive weight vector. Based on the statistical parameters of the literature association map and the real-time parameters of the environment matrix, the importance of dimensions is adjusted by nonlinear enhancement coefficients. This avoids the overwhelming of long-tail literature caused by a single dimension, such as sorting by citation count alone, while improving the distinguishability of medium-scoring literature through mechanisms such as sinusoidal excitation, ensuring that both popular and niche high-quality literature receive reasonable exposure.

[0069] S150, based on the comprehensive weight vector and the scores of the document in the preset dimensions, a comprehensive evaluation is performed to determine the retrieval score of the document, and the search results are displayed on the user interface according to the retrieval score of the document.

[0070] In this embodiment, the comprehensive weight vector and the scoring matrices of each dimension are first linearly weighted and summed to obtain the basic score of the document. Simultaneously, user behavior data from the dynamic environment matrix (such as dwell time and click preferences) is introduced, and a non-linear function is used to adjust the weights of documents with medium scores to avoid score concentration. The final retrieval score is calculated by fusing the linear and non-linear parts using a formula, ensuring a balanced presentation of highly relevant documents (e.g., high semantic matching) and real-time needs (e.g., new documents). During user interface display, pagination and progressive rendering techniques are used to optimize performance, combined with visual design (e.g., keyword highlighting, thumbnail previews) to enhance readability. Users can further refine the results by filtering conditions (e.g., time range, document type) or by secondary sorting (e.g., citation count, download count), achieving a seamless transition from intelligent scoring to interactive exploration, ultimately providing users with an efficient and intuitive document retrieval experience.

[0071] In one embodiment of this application, a comprehensive evaluation is performed based on the comprehensive weight vector and the document's score in a preset dimension to determine the document's retrieval score, including: Based on the comprehensive weight vector and the scores of the document in the preset dimensions, the first score of the document in each preset dimension is determined; Based on a preset nonlinear enhancement coefficient, the first score is calculated to determine the second score; The retrieval score of the document is determined based on the linear sum between the first score and the second score.

[0072] In one embodiment of this application, the first score of the document in each preset dimension is determined based on the comprehensive weight vector and the document's score in preset dimensions. for:

[0073] in, Indicates the preset dimension i The corresponding dimension factor, Documents j In the i Wei's rating, i and d These represent the identifier and total number of the preset dimensions, respectively.

[0074] For example, the comprehensive weight vector is Ω=(0.4,0.3,0.3), according to the literature... j semantic score =0.8, Timeliness Score The popularity score is 0.7. If the value is 0.6, then the first score is calculated based on the comprehensive weight vector and the document's score in the preset dimension. The result is: 0.4×0.8+0.3×0.7+0.3×0.6=0.71.

[0075] In one embodiment of this application, a second score is determined by calculating the first score based on a preset nonlinear enhancement coefficient. for:

[0076] in, Indicates the preset dimension k The corresponding dimension factor, Documents j In the k Wei's rating, k and d These represent the total number of identifiers for the preset dimensions.

[0077] In one embodiment of this application, the document is determined based on the linear sum between the first score and the second score. j Search rating for:

[0078] in, This represents the nonlinear enhancement coefficient determined experimentally. The above calculation process achieves nonlinear fusion and dynamic incentive of multi-dimensional scoring, ensuring that the ranking results not only conform to user intent but also possess real-time performance and personalization.

[0079] In one embodiment of this application, the search results are displayed on the user interface based on the search score of the document, including: Based on the search scores of the documents, the documents are sorted to generate a sorted list; Search results are displayed on the user interface based on the sorted list.

[0080] In one embodiment of this application, the documents are sorted in descending order based on their search scores to generate a sorted list. For example, if document 1 scores... =0.89, score of reference 2 =0.72, so the order is that reference 1 comes before reference 2.

[0081] During the display of search results, performance is optimized using pagination and progressive rendering techniques based on the structured data of the sorted list. Meanwhile, visual design (such as highlighted keywords and thumbnail previews) enhances readability. The interface layout supports responsive adjustments to adapt to different device screens. Users can further refine results by filtering criteria (such as time range or document type) or by secondary sorting (such as by citation count or download count), achieving a seamless transition from intelligent sorting to interactive exploration, ultimately providing users with an efficient and intuitive document retrieval experience.

[0082] The above process optimizes the user experience through non-linear scoring and interactive display. The non-linear enhancement coefficient performs a secondary adjustment on the linear weighted score, avoiding score concentration and making the ranking results more in line with the user's intuitive perception of "relevance." The user interface reduces information overload through technologies such as pagination and keyword highlighting, while supporting secondary filtering, such as by year and document type, and dynamic sorting, such as switching between "latest" and "most relevant" modes, achieving a seamless transition from intelligent recommendation to user-initiated exploration, ultimately improving search efficiency and satisfaction.

[0083] In this application's technical solution, environmental parameters of the search terminal, user-input query statements, and historical search records are obtained. A dynamic environment matrix is ​​generated based on the environmental parameters, and an intent vector is generated based on the query statements and historical search records. A knowledge graph is constructed based on library metadata, and document feature vectors are extracted from the knowledge graph. A document association graph is generated by linearly processing the intent vector and the document feature vectors. A comprehensive weight vector is generated based on statistical parameters extracted from the document association graph and the dynamic environment matrix. A comprehensive evaluation is performed based on the comprehensive weight vector and the document's score in a preset dimension to determine the document's search score. Based on the document's search score, the search results are displayed on the user interface. This application's technical solution, through the synergy of environmental awareness, semantic modeling, dynamic weighting, and nonlinear processing, achieves a high degree of adaptation between search results and user needs and scenarios, significantly improving the accuracy and efficiency of intelligent library retrieval.

[0084] The following describes embodiments of the intelligent library search and sorting system of this application, which can be used to execute the intelligent library search and sorting method in the above embodiments of this application. It is understood that the intelligent library search and sorting system can be a computer program (including program code) running on a computer device, for example, the intelligent library search and sorting system is an application software; the intelligent library search and sorting system can be used to execute the corresponding steps in the method provided in the embodiments of this application. For details not disclosed in the embodiments of the intelligent library search and sorting system of this application, please refer to the embodiments of the intelligent library search and sorting method described above.

[0085] Figure 3 A block diagram of a library intelligent search and sorting system according to an embodiment of this application is shown.

[0086] Reference Figure 3 As shown, a library intelligent search and sorting system according to an embodiment of this application includes: The acquisition unit 310 is used to acquire the environmental parameters of the search terminal, as well as the query statement and historical search records entered by the user; Intent unit 320 is used to generate a dynamic environment matrix based on the environment parameters and to generate an intent vector based on the query statement and the historical search records; Document unit 330 is used to construct a knowledge graph based on library metadata, extract document feature vectors from the knowledge graph, and generate a document association graph by linearly processing the intent vector and the document feature vectors. Weighting unit 340 is used to generate a comprehensive weight vector based on the statistical parameters extracted from the document association map and the dynamic environment matrix; The display unit 350 is used to perform a comprehensive evaluation based on the comprehensive weight vector and the scores of the document in a preset dimension, determine the retrieval score of the document, and display the search results on the user interface according to the retrieval score of the document.

[0087] In this application, based on the aforementioned scheme, obtaining the environmental parameters of the search terminal, the query statement input by the user, and historical search records includes: retrieving the environmental parameters of the search terminal based on the identifier of the search terminal; obtaining the query statement input by the user when a query command triggered by the user is detected; and retrieving the user's historical search records from the database based on the user's account identifier.

[0088] In this application, based on the aforementioned scheme, generating an intent vector based on the query statement and the historical retrieval records includes: performing word segmentation on the query statement to generate query words and assembling a first sequence; performing vectorization on the historical retrieval records to generate a second sequence corresponding to the historical behavior; capturing the contextual relationships in the first sequence through an encoder to generate word vectors of a preset dimension; extracting flow information in the second sequence through a gating model to generate a hidden state vector; and combining the word vectors and the hidden state vectors to generate the intent vector.

[0089] In this application, based on the aforementioned scheme, the step of constructing a knowledge graph based on library metadata and extracting document feature vectors from the knowledge graph, and generating a document association graph by linearly processing the intent vector and the document feature vectors, includes: constructing a knowledge graph based on library metadata and extracting document feature vectors from the knowledge graph; generating intent neighborhoods based on the intent vectors; determining a document set consisting of intent files related to user intents based on the relationship parameters between the intent vectors, the intent neighborhoods, and the document feature vectors; and constructing a document association graph based on the intent nodes formed by the intent vectors and the document nodes in the document set.

[0090] In this application, based on the aforementioned scheme, the step of generating a comprehensive weight vector based on the statistical parameters extracted from the document association graph and the dynamic environment matrix includes: extracting statistical parameters of a preset dimension from the document association graph; determining the dimension importance coefficient corresponding to the preset dimension based on the statistical parameters; determining the dimension factor corresponding to the preset dimension based on the statistical parameters, the dimension importance coefficient, and the dynamic environment matrix; and combining the dimension factors corresponding to the preset dimension to generate a comprehensive weight vector.

[0091] In this application, based on the aforementioned scheme, the step of comprehensively evaluating the document based on the comprehensive weight vector and the document's score in a preset dimension to determine the document's retrieval score includes: determining a first score for the document in each preset dimension based on the comprehensive weight vector and the document's score in the preset dimension; calculating a second score based on the first score using a preset nonlinear enhancement coefficient; and determining the document's retrieval score based on the linear sum between the first score and the second score.

[0092] In this application, based on the aforementioned scheme, determining the first score of the document in each preset dimension based on the comprehensive weight vector and the document's score in the preset dimensions includes: determining the first score of the document in each preset dimension based on the comprehensive weight vector and the document's score in the preset dimensions using the following formula. for:

[0093] in, Indicates the preset dimension i The corresponding dimension factor, Documents j In the i Wei's rating, i and d These represent the identifier and total number of the preset dimensions, respectively.

[0094] In this application, based on the aforementioned scheme, the step of calculating the first score based on a preset nonlinear enhancement coefficient to determine the second score includes: calculating the first score based on the preset nonlinear enhancement coefficient, and determining the second score using the following formula. for:

[0095] in, Indicates the preset dimension k The corresponding dimension factor, Documents j In the k Wei's rating, k and d These represent the total number of identifiers for the preset dimensions.

[0096] In this application, based on the aforementioned scheme, determining the retrieval score of the document according to the linear sum between the first score and the second score includes: determining the document score using the following formula based on the linear sum between the first score and the second score. j Search rating for:

[0097] in, This represents the nonlinear enhancement coefficient determined experimentally.

[0098] In this application, based on the aforementioned scheme, the step of displaying search results on the user interface according to the search score of the document includes: sorting the documents according to the search score of the document to generate a sorted list; and displaying search results on the user interface based on the sorted list.

[0099] In this application's technical solution, environmental parameters of the search terminal, user-input query statements, and historical search records are obtained. A dynamic environment matrix is ​​generated based on the environmental parameters, and an intent vector is generated based on the query statements and historical search records. A knowledge graph is constructed based on library metadata, and document feature vectors are extracted from the knowledge graph. A document association graph is generated by linearly processing the intent vector and the document feature vectors. A comprehensive weight vector is generated based on statistical parameters extracted from the document association graph and the dynamic environment matrix. A comprehensive evaluation is performed based on the comprehensive weight vector and the document's score in a preset dimension to determine the document's search score. Based on the document's search score, the search results are displayed on the user interface. This application's technical solution, through the synergy of environmental awareness, semantic modeling, dynamic weighting, and nonlinear processing, achieves a high degree of adaptation between search results and user needs and scenarios, significantly improving the accuracy and efficiency of intelligent library retrieval.

[0100] Figure 4 A schematic diagram of the structure of a computer system suitable for implementing an electronic device according to an embodiment of the present application is shown.

[0101] It should be noted that the computer system of the electronic device in this embodiment is only an example and should not impose any limitations on the function and scope of use of the embodiments of this application.

[0102] In this embodiment, the computer system includes a central processing unit 401, which can perform various appropriate actions and processes based on a program stored in read-only memory 402 or a program loaded from storage section 408 into random access memory 403, such as executing the library intelligent search and sorting method described in the above embodiment. The random access memory 403 also stores various programs and data required for system operation. The central processing unit 401, read-only memory 402, and random access memory 403 are interconnected via bus 404. Input / output interface 405 is also connected to bus 404.

[0103] The following components are connected to the input / output interface 405: an input section 406 including a keyboard, mouse, etc.; an output section 407 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a LAN (Local Area Network) card, modem, etc. The communication section 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to the input / output interface 405 as needed. A removable medium 411, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 410 as needed so that computer programs read from it can be installed into the storage section 408 as needed.

[0104] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 409, and / or installed from removable medium 411. When the computer program is executed by central processing unit 401, it performs various functions defined in the system of this application.

[0105] It should be noted that the computer-readable medium shown in the embodiments of this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying a computer-readable computer program. The transmitted data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The computer program contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, etc., or any suitable combination thereof.

[0106] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. Among them, each box in the flowchart or block diagram can represent a module, program segment, or part of the code, and the above-mentioned module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0107] The units involved in the embodiments described in this application may be implemented by software or hardware, and the units described may also be set in a processor. In some cases, the names of these units do not constitute limitations on the units themselves.

[0108] According to one aspect of this application, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the methods provided in the various alternative implementations described above.

[0109] In another aspect, this application also provides a computer-readable medium, which may be included in the electronic device described in the above embodiments; or it may exist independently and not assembled into the electronic device. The computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to implement the library intelligent search and sorting method described in the above embodiments.

[0110] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to the embodiments of this application, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0111] Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, touch terminal, or network device, etc.) to execute the methods according to the embodiments of this application.

[0112] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the embodiments disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein.

[0113] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A library intelligent search and sorting method, characterized in that, include: Obtain the environmental parameters of the search terminal, as well as the user's input query and historical search records; A dynamic environment matrix is ​​generated based on the environmental parameters, and an intent vector is generated based on the query statement and the historical search records. A knowledge graph is constructed based on the library's metadata, and document feature vectors are extracted from the knowledge graph. By performing linear processing on the intent vector and the document feature vector, a document association graph is generated. Based on the statistical parameters extracted from the literature association map and the dynamic environment matrix, a comprehensive weight vector is generated; Based on the comprehensive weight vector and the scores of the documents in the preset dimensions, a comprehensive evaluation is performed to determine the retrieval score of the documents. Based on the retrieval score of the documents, the search results are displayed on the user interface.

2. The intelligent search and sorting method for libraries according to claim 1, characterized in that, Obtain the environmental parameters of the search terminal, as well as the user's input query and historical search records, including: Based on the identifier of the search terminal, retrieve the environmental parameters of the search terminal; When a query command triggered by the user is detected, the query statement entered by the user is obtained; Based on the user's account identifier, retrieve the user's historical search records from the database.

3. The intelligent library search and sorting method according to claim 1, characterized in that, Based on the query statement and the historical retrieval records, an intent vector is generated, including: The query statement is segmented into words to generate query words and form a first sequence; The historical retrieval records are vectorized to generate a second sequence corresponding to the historical behavior; The encoder captures the contextual relationships in the first sequence and generates word vectors of a preset dimension. The flow information in the second sequence is extracted using a gating model to generate a hidden state vector; The intent vector is generated by combining the word vectors and the hidden state vectors.

4. The intelligent library search and sorting method according to claim 1, characterized in that, A knowledge graph is constructed based on the library's metadata, and document feature vectors are extracted from the knowledge graph. A document association graph is generated by linearly processing the intent vector and the document feature vectors, including: A knowledge graph is constructed based on the library's metadata, and document feature vectors are extracted from the knowledge graph. An intent neighborhood is generated based on the intent vector, and a document set consisting of intent files related to the user's intent is determined based on the relationship parameters between the intent vector, the intent neighborhood, and the file feature vector. Based on the intent nodes formed by the intent vectors and the document nodes in the document set, a document association graph is constructed.

5. The intelligent library search and sorting method according to claim 1, characterized in that, Based on the statistical parameters extracted from the literature association graph and the dynamic environment matrix, a comprehensive weight vector is generated, including: Statistical parameters of a preset dimension are extracted from the document association graph, and the dimension importance coefficient corresponding to the preset dimension is determined based on the statistical parameters. Based on the statistical parameters, the dimension importance coefficients, and the dynamic environment matrix, the dimension factors corresponding to the preset dimension are determined. The dimension factors corresponding to the preset dimension are combined to generate a comprehensive weight vector.

6. The intelligent search and sorting method for libraries according to claim 1, characterized in that, Based on the comprehensive weight vector and the document's score in the preset dimensions, a comprehensive evaluation is performed to determine the document's retrieval score, including: Based on the comprehensive weight vector and the scores of the document in the preset dimensions, the first score of the document in each preset dimension is determined; Based on a preset nonlinear enhancement coefficient, the first score is calculated to determine the second score; The retrieval score of the document is determined based on the linear sum between the first score and the second score.

7. The intelligent library search and sorting method according to claim 6, characterized in that, Based on the comprehensive weight vector and the document's scores in preset dimensions, the first score of the document in each preset dimension is determined, including: Based on the comprehensive weight vector and the document's scores in the preset dimensions, the first score of the document in each preset dimension is determined by the following formula. for: in, Indicates the preset dimension i The corresponding dimension factor, Documents j In the i Wei's rating, i and d These represent the identifier and total number of the preset dimensions, respectively; Based on a preset nonlinear enhancement coefficient, the first score is calculated to determine the second score, including: The first score is calculated based on a preset nonlinear enhancement coefficient, and the second score is determined using the following formula. for: in, Indicates the preset dimension k The corresponding dimension factor, Documents j In the k Wei's rating, k Indicates the identifier of the preset dimension.

8. The intelligent search and sorting method for libraries according to claim 7, characterized in that, The retrieval score of the document is determined based on the linear sum between the first score and the second score, including: Based on the linear sum between the first score and the second score, the literature is determined using the following formula. j Search rating for: in, This represents the nonlinear enhancement coefficient determined experimentally.

9. The intelligent library search and sorting method according to any one of claims 1-8, characterized in that, Based on the search score of the aforementioned documents, the search results are displayed on the user interface, including: Based on the search scores of the documents, the documents are sorted to generate a sorted list; Search results are displayed on the user interface based on the sorted list.

10. A library intelligent search and sorting system, characterized in that, include: The acquisition unit is used to acquire the environmental parameters of the search terminal, as well as the query statement and historical search records entered by the user; The intent unit is used to generate a dynamic environment matrix based on the environment parameters and to generate an intent vector based on the query statement and the historical search records. The document unit is used to construct a knowledge graph based on the library's metadata, extract document feature vectors from the knowledge graph, and generate a document association graph by linearly processing the intent vector and the document feature vectors. The weighting unit is used to generate a comprehensive weight vector based on the statistical parameters extracted from the document association map and the dynamic environment matrix. The display unit is used to perform a comprehensive evaluation based on the comprehensive weight vector and the scores of the document in a preset dimension, determine the retrieval score of the document, and display the search results on the user interface according to the retrieval score of the document.

Images