Intelligent search engine system based on Simbert algorithm
Simbert's algorithm-based intelligent search engine system solves the accuracy and efficiency problems of traditional search engines in complex query scenarios by deeply understanding the query context and semantics, thus achieving a personalized search experience and efficient information acquisition.
Patent Information
- Application Number
- CN202410982316.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-15
AI Technical Summary
Traditional search engines struggle to accurately capture user needs when dealing with semantic ambiguity, synonym substitution, and complex query intent, leading to a decline in the relevance and accuracy of search results.
The intelligent search engine system employing the Simbert algorithm utilizes the Simbert model to deeply understand the query context and semantics through modules for data collection and preprocessing, index building, query processing, similarity calculation, and result ranking. This generates semantic vectors for matching and ranking.
It improves the relevance and accuracy of search results, can handle complex query scenarios, provides a personalized search experience, and improves search efficiency while reducing reliance on manual screening and optimization.
Smart Images

Figure CN118964524A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of intelligent search engines, and in particular to an intelligent search engine system based on a Simbert algorithm. Background Art
[0002] The intelligent search engine system based on the Simbert algorithm is a method of improving search engines using natural language processing (NLP) models. Simbert is a pre-trained Chinese language model, similar to the English BERT model, which is optimized and trained specifically for the Chinese context;
[0003] In the era of information explosion, search engines are a bridge connecting users with massive network resources, and their importance is self-evident. Traditional search engines mainly rely on keyword matching technology to determine the relevance of documents by directly comparing user query keywords with the vocabulary in the document. However, this keyword-based approach has significant limitations, especially in dealing with semantic ambiguity, synonym replacement, and complex query intent scenarios. It is often difficult to accurately capture the real needs of users, resulting in a significant reduction in the relevance and accuracy of search results. Summary of the invention
[0004] In order to solve the above technical problems, the present invention provides an intelligent search engine system based on the Simbert algorithm, which can improve semantic understanding ability, handle complex query scenarios, realize personalized search experience, and improve search efficiency and accuracy.
[0005] In a first aspect, the present invention provides an intelligent search engine system based on the Simbert algorithm, the system comprising:
[0006] The data collection and preprocessing module collects massive text data, preprocesses the collected text data, and builds a corpus;
[0007] The index building module encodes the preprocessed corpus through the Simbert algorithm, generates semantic vectors, and stores them in the search engine;
[0008] The query processing module obtains the query request input by the user to the search engine, encodes the query request through the Simbert algorithm, and generates a query vector;
[0009] The similarity calculation module matches the user's query vector with the semantic vector stored in the database, calculates the similarity between the query vector and the semantic vector, and obtains the similarity score of the query request;
[0010] The result sorting and feedback module sorts the corresponding text data according to the similarity score, obtains the best matching document set, and feeds back the best matching document set to the user.
[0011] Furthermore, the method for preprocessing the collected text data includes:
[0012] Remove duplicates from text data through hash comparison;
[0013] Remove non-text information from text data, and clean up meaningless symbols, spaces, and noise data of advertising content;
[0014] Standardize text formats, including converting full-width characters to half-width characters, standardizing number formats, standardizing date formats, and standardizing uppercase and lowercase letters;
[0015] For Chinese text, use the word segmentation tool to split the continuous text into word sequences and perform part-of-speech tagging;
[0016] Introduce a synonym dictionary to replace words in the text.
[0017] Furthermore, the method for generating the semantic vector includes:
[0018] The preprocessed corpus is used as the input of the Simbert model; the Simbert model segments the text and adds tags;
[0019] The Simbert model learns the contextual relationships in the text through a multi-layer self-attention mechanism, generating a high-dimensional vector for each input word position;
[0020] For the entire document, the Simbert model uses the output vector at a specific position as the global semantic vector of the document;
[0021] The original high-dimensional semantic vector is subjected to dimensionality reduction processing to obtain the semantic vector.
[0022] Furthermore, the method for generating the query vector includes:
[0023] When a user enters a query request on the search engine interface, the query request is parsed, including removing irrelevant characters, word segmentation, and error checking and correction of user input;
[0024] Using the Simbert model, the semantic content of user queries is analyzed to identify potential semantic associations, implicit meanings, and contextual dependencies;
[0025] Simbert encodes the parsed and understood query request and converts it into a high-dimensional semantic vector to obtain the query vector.
[0026] Furthermore, the method for obtaining the similarity score of the query request includes:
[0027] Normalize the user's query vector and the semantic vector stored in the database;
[0028] Compare the query vector with the semantic vectors of all documents in the index;
[0029] Using the cosine similarity method, the similarity between the query vector and each document vector is calculated one by one to obtain a similarity score;
[0030] For each document, record its similarity score with the query vector.
[0031] Furthermore, the method of feeding back the best matching document set to the user includes:
[0032] For each search result, display the title and a snippet summary related to the query, highlighting sentences containing the user's query keywords;
[0033] Highlight the user's query terms in the search results to help users quickly locate information;
[0034] Paginate search results, displaying a fixed number of results per page;
[0035] Allows users to re-sort results based on date, relevance criteria.
[0036] Furthermore, the data sources of the data collection and preprocessing module include web pages, news, social media, academic paper databases, and public APIs.
[0037] Furthermore, the corpus content includes web page content, academic documents, electronic books, multimedia metadata, product descriptions, service information, and problem discussion content.
[0038] In a third aspect, the present application provides an electronic device, comprising a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the transceiver, the memory, and the processor are connected via the bus, and the computer program, when executed by the processor, implements the steps of any one of the above methods.
[0039] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps in any one of the above-mentioned methods when executed by a processor.
[0040] Compared with the prior art, the present invention has the following beneficial effects: the Simbert model is pre-trained with large-scale Chinese corpus and can deeply understand the context and semantics of query sentences, rather than relying solely on keyword matching; when the system faces queries containing synonyms, near-synonyms, or semantically ambiguous queries, it can more accurately grasp the user's true intention and improve the relevance of search results;
[0041] Traditional search engines have difficulty dealing with queries that contain multiple conditions or complex logical relationships. However, the Simbert-based system, due to its deep understanding of language structure and context, can better parse such complex queries and provide more accurate search feedback.
[0042] For long-tail queries and fuzzy queries, the Simbert model can effectively match highly relevant content even when the query keywords do not appear directly in the document by capturing the deep semantic association between the query and the document, thus expanding the effective coverage of the search.
[0043] Through automated data preprocessing, index building, and query processing, the system reduces reliance on manual screening and optimization, improves search efficiency and system scalability; as a pre-trained model, the Simbert model can adapt to specific language features and professional terminology in different fields through continuous fine-tuning and training, so that the system can maintain high search quality in different application scenarios;
[0044] In summary, the intelligent search engine system based on the Simbert algorithm is significantly superior to the traditional keyword matching-based search engine by improving semantic understanding capabilities, handling complex query scenarios, achieving personalized search experience, and improving search efficiency and accuracy. It can better meet users' needs for high-quality information acquisition in the era of information explosion. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] Figure 1 It is a structural diagram of an intelligent search engine system based on the Simbert algorithm; DETAILED DESCRIPTION
[0046] In the description of the present application, those skilled in the art should know that the present application can be implemented as a method, an apparatus, an electronic device, and a computer-readable storage medium. Therefore, the present application can be specifically implemented in the following forms: complete hardware, complete software (including firmware, resident software, microcode, etc.), a combination of hardware and software. In addition, in some embodiments, the present application can also be implemented in the form of a computer program product in one or more computer-readable storage media, and the computer-readable storage medium contains computer program code.
[0047] The above-mentioned computer-readable storage medium may adopt any combination of one or more computer-readable storage media. Computer-readable storage media include: electrical, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or devices, or any combination of the above. More specific examples of computer-readable storage media include: portable computer disks, hard disks, random access memories, read-only memories, erasable programmable read-only memories, flash memories, optical fibers, optical disc read-only memories, optical storage devices, magnetic storage devices, or any combination of the above. In the present application, computer-readable storage media can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device, or device.
[0048] The acquisition, storage, use, and processing of data in the technical solution of this application comply with the relevant provisions of national laws.
[0049] The present application describes the provided methods, devices, and electronic devices through flowcharts and / or block diagrams.
[0050] It should be understood that each box in the flowchart and / or block diagram and the combination of boxes in the flowchart and / or block diagram can be implemented by computer-readable program instructions. These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer or other programmable data processing device to produce a machine, and these computer-readable program instructions are executed by a computer or other programmable data processing device to produce a device that implements the functions / operations specified by the boxes in the flowchart and / or block diagram.
[0051] These computer-readable program instructions may also be stored in a computer-readable storage medium that enables a computer or other programmable data processing device to work in a specific manner. In this way, the instructions stored in the computer-readable storage medium produce an instruction device product including functions / operations specified in the blocks in the flowchart and / or block diagram.
[0052] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device, so that a series of operational steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby enabling the instructions executed on the computer or other programmable data processing apparatus to provide a process for implementing the functions / operations specified in the blocks in the flowchart and / or block diagram.
[0053] The present application is described below in conjunction with the drawings in the present application.
[0054] Embodiment 1: Figure 1 As shown, an intelligent search engine system based on the Simbert algorithm of the present invention specifically includes the following steps:
[0055] The data collection and preprocessing module collects massive text data, preprocesses the collected text data, and builds a corpus;
[0056] The index building module encodes the preprocessed corpus through the Simbert algorithm, generates semantic vectors, and stores them in the search engine;
[0057] The query processing module obtains the query request input by the user to the search engine, encodes the query request through the Simbert algorithm, and generates a query vector;
[0058] The similarity calculation module matches the user's query vector with the semantic vector stored in the database, calculates the similarity between the query vector and the semantic vector, and obtains the similarity score of the query request;
[0059] The result sorting and feedback module sorts the corresponding text data according to the similarity score, obtains the best matching document set, and feeds back the best matching document set to the user.
[0060] The Simbert model has been pre-trained with a large amount of Chinese corpus and can deeply understand the context and semantics of query statements, rather than relying solely on keyword matching. This enables the system to more accurately grasp the user's true intention when faced with queries containing synonyms, near-synonyms, or semantically ambiguous words, thereby improving the relevance of search results.
[0061] Traditional search engines have difficulty dealing with queries that contain multiple conditions or complex logical relationships. However, the Simbert-based system, due to its deep understanding of language structure and context, can better parse such complex queries and provide more accurate search feedback.
[0062] For long-tail queries and fuzzy queries, the Simbert model can effectively match highly relevant content even when the query keywords do not appear directly in the document by capturing the deep semantic association between the query and the document, thus expanding the effective coverage of the search.
[0063] Through automated data preprocessing, index building, and query processing, the system reduces reliance on manual screening and optimization, improves search efficiency and system scalability; as a pre-trained model, the Simbert model can adapt to specific language features and professional terminology in different fields through continuous fine-tuning and training, so that the system can maintain high search quality in different application scenarios;
[0064] In summary, the intelligent search engine system based on the Simbert algorithm significantly outperforms traditional keyword-matching-based search engines by enhancing semantic understanding capabilities, handling complex query scenarios, achieving personalized search experiences, and improving search efficiency and accuracy. It can better meet users' high-quality information acquisition needs in the era of information explosion.
[0065] Regarding the data collection and preprocessing module, specifically:
[0066] Clarify the goals of data collection, which involves formulating web crawler strategies; data sources include web pages, news, social media, academic paper databases, public APIs, etc.; the selected data sources should be widely covered and relevant to the theme of the search engine to ensure the comprehensiveness and diversity of the collected information;
[0067] According to the determined data sources, use existing web crawler tools to automatically crawl web page content;
[0068] There will be duplicate content in the data collected from multiple sources. Remove duplicate items through hash comparison or other deduplication techniques to maintain the purity of the corpus;
[0069] Remove non-text information such as HTML tags, CSS styles, JavaScript code, etc., and at the same time clean up noise data such as meaningless symbols, spaces, and advertising content;
[0070] Unify the text format, such as converting full-width characters to half-width, standardizing number and date formats, and unifying case, etc., for subsequent processing;
[0071] For Chinese text, use a word segmentation tool to split continuous text into a sequence of words, and perform part-of-speech tagging if necessary to prepare for subsequent vectorization processing;
[0072] Remove common stop words such as "de", "he", "zai", etc. These words contribute little to understanding the meaning of the text in most cases. Removing them can reduce noise and improve processing efficiency;
[0073] To enhance the model's understanding of synonymous expressions, a thesaurus can be introduced to replace the vocabulary in the text to a certain extent and increase the richness of the corpus;
[0074] The cleaned and processed text data needs to be organized and stored in an efficient data structure; at the same time, construct a document-term matrix to obtain a corpus for subsequent model training or directly for information retrieval;
[0075] The content of the corpus includes:
[0076] Web content: various web pages scraped from the Internet, including news articles, blogs, forum posts, social media updates, etc., covering a wide range of topics and fields;
[0077] Documents and e-books: including academic papers, reports, e-books and other professional documents, which have high information value and authority;
[0078] Multimedia metadata: Text information such as titles, descriptions, and tags associated with images and videos is also part of the corpus;
[0079] Product and service information: including product descriptions and service descriptions of the e-commerce platform;
[0080] Question and discussion content: User-generated content such as questions and answers, comments, and forum discussions on question-and-answer websites can reflect actual language usage and enhance the model's ability to understand everyday language;
[0081] The data collection and preprocessing module plays a key role in the intelligent search engine system based on the Simbert algorithm. Through effective data collection and preprocessing, the system can establish a rich and high-quality corpus, providing a reliable foundation for subsequent query processing and similarity calculation.
[0082] For the index building module, specifically:
[0083] The preprocessed corpus is used as the input of the Simbert model. The Simbert model will first tokenize the text and add specific tags to match the model's input format.
[0084] The Simbert model learns the contextual relationships in the text through a multi-layer self-attention mechanism, generating a high-dimensional vector for each input word position. These vectors comprehensively consider the contextual information of the words, thereby capturing the deep semantics of the sentence.
[0085] For the entire document or text block, the Simbert model uses the output vector at a specific position as the global semantic vector of the document; this vector can be regarded as a compact representation of the document, which contains the main semantic information of the document;
[0086] In order to reduce storage space and speed up retrieval, the original high-dimensional semantic vector needs to be reduced in dimensionality. Commonly used techniques include PCA, t-SNE, or the use of special quantization methods;
[0087] The reduced semantic vectors and the corresponding document IDs are stored in an efficient index structure, which facilitates the subsequent rapid search for documents that are most similar to a specific query vector.
[0088] The index building module converts text information into vector representation in a high-dimensional semantic space through the Simbert model, and manages these vectors through efficient data structures and algorithms, providing a basis for subsequent query processing and similarity calculation.
[0089] For the query processing module, specifically:
[0090] When a user enters a query request in the search engine interface, the query processing module first receives and parses the request, including removing irrelevant characters, word segmentation, and basic error checking and correction of user input;
[0091] Using the Simbert model, the query processing module deeply analyzes the semantic content of user queries. With its deep Chinese language understanding ability, Simbert not only matches the surface vocabulary, but also understands the intention behind the query and identifies the potential semantic associations, implicit meanings and contextual dependencies.
[0092] Simbert encodes the parsed and understood query request and converts it into a high-dimensional semantic vector. Through the multi-layer self-attention and feedforward neural network in the Transformer architecture, it captures the keywords in the query and their relationships to form a vector that can represent the query semantics. This vector can effectively express the deep meaning of the query, surpassing the limitations of traditional keyword-based matching.
[0093] The query processing module is designed to be highly responsive and flexible, capable of processing user query requests in real time, and can remain efficient even in the face of a large number of concurrent requests. In addition, to further enhance the user experience, the module will use a caching strategy to speed up the response speed of frequent queries, and use algorithm optimization to ensure that query processing is both fast and accurate.
[0094] Once the query vector is generated, the query processing module will pass it to the similarity calculation module for comparison with the stored document semantic vectors to find the most relevant documents;
[0095] In summary, the query processing module transforms the user's natural language query requests into machine-processable semantic vectors through the powerful learning ability and semantic understanding function of the Simbert model, which greatly improves the search accuracy and user satisfaction.
[0096] For the similarity calculation module, specifically:
[0097] With the support of the Simbert algorithm, both the user's query request and the document content are converted into high-dimensional semantic vectors; these vectors are located in a multidimensional space, and each dimension represents a specific semantic feature; similarity calculation is first based on this vector space model, and their similarity is measured by calculating the distance or angle between two vectors;
[0098] Cosine similarity is one of the similarity measurement methods, which is applicable to high-dimensional sparse vectors. It evaluates the similarity between two vectors by calculating the cosine value of the angle between them. The closer the cosine value is to 1, the more consistent the directions of the two vectors are, that is, the higher the semantic relevance between the query and the document.
[0099] In more complex search engine systems, in addition to text information, multimedia information such as images and audio is also included; the similarity calculation module needs to consider how to fuse the feature vectors of different modalities to improve the comprehensiveness and accuracy of the search;
[0100] For each user's query request, it is converted into a semantic vector through the Simbert model;
[0101] In order to eliminate the influence of vector length, the vector is normalized to ensure the validity of cosine similarity;
[0102] Compare this query vector with the semantic vectors of all documents in the previously built index;
[0103] Using the cosine similarity method, the similarity between the query vector and each document vector is calculated one by one;
[0104] Record the similarity score with the query vector for each document;
[0105] The documents are ranked according to the calculated similarity scores, and the documents with the highest scores are considered to be most relevant to the query;
[0106] The similarity calculation module uses advanced vector space models and effective similarity measurement methods to accurately quantify the degree of semantic matching between queries and documents based on the semantic vectors generated by Simbert, providing users with a more accurate and personalized search experience.
[0107] For the result sorting and feedback module, specifically:
[0108] Preliminary sorting based on similarity scores, using the cosine similarity or other related metrics calculated between the query vector and each document vector in the corpus as the basic sorting criterion; however, this is only the first step in sorting, because a single similarity score may not be enough to fully reflect the quality and applicability of a document; therefore, other factors such as the authority, timeliness, and user behavior signals of the document are also considered to form a multi-dimensional sorting model;
[0109] In order to more accurately predict user satisfaction with search results, machine learning methods are used, such as gradient boosting trees, deep neural networks, or learning ranking models such as LambdaMART. These models can learn which features are more important for ranking through historical user behavior data, and automatically adjust the weight of each feature to achieve more intelligent ranking.
[0110] Taking into account the diversity and personalization of user needs, the system will use user profiling technology to adjust the order of search results based on the user's search history, interest preferences, geographic location and other information to provide content that is more in line with personal preferences;
[0111] Methods for feeding back the best matching document set to the user include:
[0112] Summary generation: For each search result, the title and the most relevant snippet summary are displayed, and the text is further processed to highlight sentences containing the user's query keywords;
[0113] Highlighting: Highlighting the user's query terms in the search results to help users quickly locate information;
[0114] Pagination processing: If there are many search results, pagination processing is performed, and a fixed number of results are displayed on each page to optimize the user experience;
[0115] Provide filtering and sorting options: Allow users to re-sort or filter results based on criteria such as date, relevance, etc.
[0116] User feedback mechanism: Provide a mechanism for users to report irrelevant results or give feedback on the quality of results for further training and optimization of the model;
[0117] The result ranking and feedback module uses complex algorithms and models, combined with users' real-time feedback and long-term behavior patterns, to dynamically adjust the ranking of search results, aiming to maximize user satisfaction and search efficiency. This process not only relies on the high-quality semantic understanding capabilities provided by the Simbert algorithm, but also requires a highly flexible and intelligent learning mechanism to ensure that the search engine can continue to evolve and better adapt to user needs and changes in network content.
[0118] The present application also provides an electronic device, including a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor. The transceiver, the memory, and the processor are respectively connected via the bus. When the computer program is executed by the processor, each process of the above-mentioned method for controlling output data is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.
[0119] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. An intelligent search engine system based on the Simbert algorithm, It is characterized in that The system comprises: The data collection and preprocessing module collects massive text data, preprocesses the collected text data, and builds a corpus; The index building module encodes the preprocessed corpus through the Simbert algorithm, generates semantic vectors, and stores them in the search engine; The query processing module obtains the query request input by the user to the search engine, encodes the query request through the Simbert algorithm, and generates a query vector; The similarity calculation module matches the user's query vector with the semantic vector stored in the database, calculates the similarity between the query vector and the semantic vector, and obtains the similarity score of the query request; The result sorting and feedback module sorts the corresponding text data according to the similarity score, obtains the best matching document set, and feeds back the best matching document set to the user.
2. An intelligent search engine system based on the Simbert algorithm as claimed in claim 1, It is characterized in that Methods for preprocessing the collected text data include: Remove duplicates from text data through hash comparison; Remove non-text information from text data, and clean up meaningless symbols, spaces, and noise data of advertising content; Standardize text formats, including converting full-width characters to half-width characters, standardizing number formats, standardizing date formats, and standardizing uppercase and lowercase letters; For Chinese text, use the word segmentation tool to split the continuous text into word sequences and perform part-of-speech tagging; Introduce a synonym dictionary to replace words in the text.
3. An intelligent search engine system based on the Simbert algorithm as claimed in claim 1, It is characterized in that The method for generating the semantic vector includes: The preprocessed corpus is used as the input of the Simbert model; the Simbert model segments the text and adds tags; The Simbert model learns the contextual relationships in the text through a multi-layer self-attention mechanism, generating a high-dimensional vector for each input word position; For the entire document, the Simbert model uses the output vector at a specific position as the global semantic vector of the document; The original high-dimensional semantic vector is subjected to dimensionality reduction processing to obtain the semantic vector.
4. An intelligent search engine system based on the Simbert algorithm as claimed in claim 1, It is characterized in that The method for generating the query vector includes: When a user enters a query request on the search engine interface, the query request is parsed, including removing irrelevant characters, word segmentation, and error checking and correction of user input; Using the Simbert model, the semantic content of user queries is analyzed to identify potential semantic associations, implicit meanings, and contextual dependencies; Simbert encodes the parsed and understood query request and converts it into a high-dimensional semantic vector to obtain the query vector.
5. An intelligent search engine system based on the Simbert algorithm as claimed in claim 1, It is characterized in that Methods for obtaining the similarity score of a query request include: Normalize the user's query vector and the semantic vector stored in the database; Compare the query vector with the semantic vectors of all documents in the index; Using the cosine similarity method, the similarity between the query vector and each document vector is calculated one by one to obtain a similarity score; For each document, record its similarity score with the query vector.
6. An intelligent search engine system based on the Simbert algorithm as claimed in claim 1, It is characterized in that Methods for feeding back the best matching document set to the user include: For each search result, display the title and a snippet summary related to the query, highlighting sentences containing the user's query keywords; Highlight the user's query terms in the search results to help users quickly locate information; Paginate search results, displaying a fixed number of results per page; Allows users to re-sort results based on date, relevance criteria.
7. An intelligent search engine system based on the Simbert algorithm as claimed in claim 1, It is characterized in that The data sources of the data collection and preprocessing module include web pages, news, social media, academic paper databases, and public APIs.
8. An intelligent search engine system based on the Simbert algorithm as claimed in claim 1, It is characterized in that The corpus content includes web page content, academic documents, e-books, multimedia metadata, product descriptions, service information, and problem discussion content.
9. An intelligent search engine based on the Simbert algorithm, comprising a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the transceiver, the memory, and the processor are connected via the bus. It is characterized in that When the computer program is executed by the processor, the steps in the method according to any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps in the method according to any one of claims 1 to 7 are implemented.
Citation Information
Cited By
Data search method and system based on artificial intelligence
CN119227013A
Media file querying and calling method and system based on AI dialogue scene
CN119474419A
Term data intelligent management system based on machine learning
CN119513289A
Multi-platform store normalization method
CN119807399A
Event context reduction method and system based on vector retrieval and large language model
CN120780759A
