ES-based support vector intelligent information retrieval and analysis system

Abstract

The invention relates to information retrieval analysis, in particular to an ES-based support vector intelligent information retrieval analysis system, which comprises a data acquisition module for collecting data from various data sources, preprocessing the collected data, ensuring the quality and consistency of the data, and sending the preprocessed data to a vector conversion module; the vector conversion module is used for converting different types of data into vector representation, making adaptive vector conversion methods for different types of data to ensure that vectors can accurately reflect data features, and sending the vectors to the Elasticsearch storage and index module; the Elasticsearch storage and index module is used for storing the vectors into an Elasticsearch database and optimizing an index algorithm according to the characteristics of the vectors; according to the technical scheme provided by the invention, the defect that useful information is difficult to accurately and quickly obtain from mass data in the prior art can be effectively overcome.

Description

Technical Field

[0001] The present invention relates to information retrieval analysis, and in particular to an intelligent information retrieval analysis system based on support vectors of ES. Background Art

[0002] With the rapid development of information technology, the amount of data is growing explosively. How to accurately and quickly obtain useful information from massive data has become a pressing issue. Traditional information retrieval and analysis methods have certain limitations in dealing with complex data types and semantic understanding. As a powerful open source search engine, Elasticsearch has performed well in text search and other aspects, but there is still much room for improvement in processing vector data and vector-based deep semantic analysis and intelligent information retrieval.

[0003] At present, the processing and application of vector data often rely on specialized vector databases or complex machine learning frameworks, which increases the complexity and cost of the system. At the same time, the existing information retrieval system is difficult to meet the growing user needs in terms of accuracy, real-time and personalization. Therefore, it is necessary to develop an intelligent information retrieval and analysis system based on ES support vectors to improve the quality and efficiency of information retrieval and analysis. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the above-mentioned shortcomings of the prior art, the present invention provides an intelligent information retrieval and analysis system based on ES support vectors, which can effectively overcome the defect of the prior art that it is difficult to accurately and quickly obtain useful information from massive data.

[0006] (II) Technical solution

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0008] An intelligent information retrieval and analysis system based on ES support vectors, including a data acquisition module, a vector conversion module, an Elasticsearch storage and indexing module, a vector reasoning module, and a user interaction module;

[0009] The data acquisition module collects data from various data sources, pre-processes the collected data to ensure the quality and consistency of the data, and sends the pre-processed data to the vector conversion module;

[0010] The vector conversion module converts different types of data into vector representations. By formulating adaptive vector conversion methods for different types of data, it ensures that the vectors can accurately reflect the data characteristics and sends the vectors to the Elasticsearch storage and indexing module.

[0011] The Elasticsearch storage and indexing module stores vectors in the Elasticsearch database and optimizes the indexing algorithm based on the characteristics of vectors to improve the storage efficiency and retrieval speed of vectors. It also supports dynamic index updates to ensure that the system can respond to query requests accurately and quickly.

[0012] The vector reasoning module implements various vector reasoning operations based on the vector computing function of Elasticsearch to mine the potential relationships between vectors, discover potential patterns and anomalies of vectors, and provide users with valuable information analysis results.

[0013] The user interaction module provides users with a friendly and convenient front-end interactive interface, supports users to input query requests in a variety of ways, calls corresponding retrieval functions according to query requests, and displays information retrieval results in an intuitive way.

[0014] Preferably, the data acquisition module collects data from various data sources, pre-processes the collected data to ensure the quality and consistency of the data, and sends the pre-processed data to the vector conversion module, including:

[0015] S11. According to the connection protocols and interfaces of different data sources, corresponding connection parameters are set to establish a stable data collection channel for data collection;

[0016] S12. Perform corresponding preprocessing operations based on the characteristics of different types of data:

[0017] 1) For text data, remove stop words and punctuation marks, perform stem extraction, part-of-speech tagging and word segmentation;

[0018] 2) For image data, perform image scaling, cropping, and normalization operations so that the image size and pixel value meet the requirements of subsequent processing;

[0019] 3) For audio data, perform audio noise reduction and segmentation processing to improve the quality and processability of audio data;

[0020] S13, sending the preprocessed data to the vector conversion module;

[0021] Among them, data sources include relational databases, non-relational databases, file systems and network data sources.

[0022] Preferably, the vector conversion module converts different types of data into vector representations, including:

[0023] 1) For text data, use the word embedding model to convert words into corresponding vectors, and then use the document vector generation algorithm to combine the vectors corresponding to the words into a text feature vector representing the entire document;

[0024] 2) For image data, the image data is input into the convolutional neural network CNN, which is initialized with pre-trained weights. After multiple layers of convolution and pooling operations, the output of the last fully connected layer is extracted as the image feature vector;

[0025] 3) For audio data, the audio data is input into the recurrent neural network RNN ​​to obtain the audio feature vector, and the audio processing library is used to assist in the processing of the audio data.

[0026] Preferably, the Elasticsearch storage and indexing module stores the vectors in the Elasticsearch database, optimizes the indexing algorithm according to the characteristics of the vectors to improve the storage efficiency and retrieval speed of the vectors, and supports dynamic index updates to ensure that the system can accurately and quickly respond to query requests, including:

[0027] S31. Install and configure the Elasticsearch database and create an index structure specifically for storing vectors:

[0028] For vectors, define the appropriate field type and set the index structure parameters based on the vector dimension and similarity function;

[0029] S32, importing the vectors sent by the vector conversion module into the Elasticsearch database in batches through the Elasticsearch API interface, and adopting an optimized storage strategy during the import process;

[0030] S33. Regularly optimize and update the index structure to cope with data changes and growth, maintain the stability of retrieval speed, and ensure that the system can respond to query requests accurately and quickly.

[0031] Preferably, the vector reasoning module implements a variety of vector reasoning operations based on the vector calculation function of Elasticsearch to mine potential relationships between vectors and discover potential patterns and abnormal situations of vectors, including:

[0032] 1) Use clustering algorithms to group vectors. By analyzing the characteristics and distribution of vectors, similar vectors are classified into the same category to explore the potential relationships between vectors and discover the potential patterns of vectors.

[0033] 2) By setting a threshold or using anomaly detection algorithms to monitor vectors in real time, abnormal vectors that differ greatly from normal vector patterns can be identified.

[0034] Preferably, the vector reasoning module monitors data changes and user behaviors in real time, dynamically adjusts the vector model and index structure, and automatically triggers the vector conversion module to update the vectors and the Elasticsearch storage and indexing module to update the index structure when new data is added or the user's query behavior changes. At the same time, the parameters of the vector reasoning algorithm are dynamically adjusted according to real-time data and user feedback to improve the real-time and accuracy of retrieval and analysis.

[0035] Preferably, the user interaction module provides users with a friendly and convenient front-end interactive interface, supports users to input query requests in a variety of ways, calls corresponding search functions according to the query requests, and displays information retrieval results in an intuitive manner, including:

[0036] S51, using a front-end development framework to build a front-end interactive interface, where the user inputs a query request in a query input box by entering text, uploading an image, or uploading an audio;

[0037] S52, the front-end interactive interface sends the query request to the back-end server, and the back-end server calls the vector reasoning module to perform data retrieval and returns the information retrieval result to the front-end interactive interface;

[0038] S53, after receiving the information retrieval result, the front-end interactive interface presents the information retrieval result to the user in a corresponding form according to the data type and result characteristics;

[0039] Among them, the front-end interactive interface includes a query input box, a query button and a result display area.

[0040] Preferably, after being called by the backend server, the vector reasoning module performs data retrieval, including:

[0041] S521, performing semantic analysis on the query request, using the knowledge graph and word vector model to expand the semantic scope of the query keyword, and generating a semantic vector;

[0042] S522, fusing the semantic vector with the original vector corresponding to the query request to generate a query vector;

[0043] S523. Based on the vector calculation function and similarity function provided by Elasticsearch, the similarity between the query vector and the vectors stored in the Elasticsearch database is calculated, sorted according to the similarity, and the most similar result is returned to the backend server.

[0044] (III) Beneficial effects

[0045] Compared with the prior art, the intelligent information retrieval and analysis system based on ES support vector provided by the present invention has the following beneficial effects:

[0046] 1) Improve the quality of information retrieval

[0047] A. Deep semantic understanding

[0048] By developing adaptive vector conversion methods for different types of data, such as converting text, images, audio and other data into vector representations, we can deeply explore the semantic information behind the data. For text retrieval, it is no longer limited to simple keyword matching, but is based on the semantic information contained in word vectors and document vectors. Even if the keywords entered by the user are not completely consistent with the expressions in the literature, as long as the semantics are similar, the relevant literature can be accurately retrieved, which greatly improves the accuracy of information retrieval;

[0049] B. Multimodal Information Fusion Retrieval

[0050] The system can process multiple types of data at the same time, realizing multi-modal information fusion retrieval. For example, in news media applications, users can obtain comprehensive and relevant news reports by inputting a text describing a news event, a related image, and a live audio clip. This breaks through the limitations of traditional single-modal information retrieval and provides users with more accurate and richer information retrieval results.

[0051] 2) Enhance information analysis capabilities

[0052] A. Accurate Data Clustering and Pattern Discovery

[0053] Using vector clustering algorithms for analysis can automatically discover hidden patterns and structures in data. For example, in the field of market research, consumers' purchasing behavior data (such as purchase information, purchase time, and consumption amount) can be converted into vector representations and then clustered. This can clearly divide different consumer groups, as well as the consumption characteristics and preference patterns of each consumer group, providing a strong basis for companies to formulate marketing strategies.

[0054] B. Real-time abnormality monitoring

[0055] Vector-based anomaly monitoring can monitor data changes in real time and detect abnormal behaviors in a timely manner. For example, in financial transaction scenarios, transaction data (such as transaction amount, transaction time, and transaction object) can be converted into vector representations for real-time monitoring, which can quickly identify possible fraudulent transactions or abnormal transactions, effectively prevent financial risks, and ensure the safe and stable operation of the financial system.

[0056] 3) Optimize user experience

[0057] A. Personalized service realization

[0058] The vector reasoning module automatically triggers vector updates and index structure updates based on data and user behavior changes, and can provide personalized services for users. For example, in social networks, the system generates user vectors based on users' browsing history, likes, comments, and friend relationships, and then recommends content that highly matches their interests and hobbies and friends who may be interested in them, thereby enhancing users' stickiness and satisfaction with the platform.

[0059] B. Friendly interactive interface and convenient query

[0060] The user interaction module supports users to input query requests in a variety of ways, whether it is inputting text, uploading images or voice input, which can easily perform queries. For example, in a smart home environment, users can describe "play cheerful music" through voice, or upload a picture of the home decoration they want to search for, and the system can accurately understand the user's intention and quickly give results;

[0061] 4) System performance advantages

[0062] A. Efficient storage and fast retrieval

[0063] The optimized Elasticsearch vector storage and index structure make vector storage more efficient. At the same time, the specially designed indexing algorithm and vector inference algorithm ensure the speed of information retrieval. Even when processing massive data, query requests can be responded to in a short time, meeting users' real-time requirements.

[0064] B. Reduce system complexity and cost

[0065] The intelligent information retrieval and analysis system based on ES's support vector implements vector support, avoiding the introduction of additional complex vector databases or machine learning frameworks. The open source and maturity of Elasticsearch reduce the cost of system development, maintenance, and expansion, and the unified processing framework can process multiple types of data at the same time, improving the utilization of system resources and effectively solving the integration problem between different types of data processing systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0067] Figure 1 It is a system schematic diagram of the present invention. DETAILED DESCRIPTION

[0068] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0069] An intelligent information retrieval and analysis system based on ES support vector, such as Figure 1 As shown, it includes a data acquisition module, a vector conversion module, an Elasticsearch storage and indexing module, a vector reasoning module and a user interaction module;

[0070] The data acquisition module collects data from various data sources, pre-processes the collected data to ensure the quality and consistency of the data, and sends the pre-processed data to the vector conversion module;

[0071] The vector conversion module converts different types of data into vector representations. By formulating adaptive vector conversion methods for different types of data, it ensures that the vectors can accurately reflect the data characteristics and sends the vectors to the Elasticsearch storage and indexing module.

[0072] The Elasticsearch storage and indexing module stores vectors in the Elasticsearch database and optimizes the indexing algorithm based on the characteristics of vectors to improve the storage efficiency and retrieval speed of vectors. It also supports dynamic index updates to ensure that the system can respond to query requests accurately and quickly.

[0073] The vector reasoning module implements various vector reasoning operations based on the vector computing function of Elasticsearch to mine the potential relationships between vectors, discover potential patterns and anomalies of vectors, and provide users with valuable information analysis results.

[0074] The user interaction module provides users with a friendly and convenient front-end interactive interface, supports users to input query requests in a variety of ways, calls corresponding retrieval functions according to query requests, and displays information retrieval results in an intuitive way.

[0075] ① The data acquisition module collects data from various data sources, pre-processes the collected data to ensure the quality and consistency of the data, and sends the pre-processed data to the vector conversion module, including:

[0076] S11. According to the connection protocols and interfaces of different data sources, corresponding connection parameters are set to establish a stable data collection channel for data collection (for example, for a relational database, connection is performed through the corresponding database driver and connection string; for a network data source, web page data is obtained through a web crawler, and crawling rules, starting URL, depth limit and other parameters are set);

[0077] S12. Perform corresponding preprocessing operations based on the characteristics of different types of data:

[0078] 1) For text data, remove stop words and punctuation, perform stemming, part-of-speech tagging, and word segmentation (for example, when processing English text data, remove stop words such as "is", "a", and "the", and uniformly treat words such as "running" and "runs" as "run", perform stemming, and tag their parts of speech);

[0079] 2) For image data, perform image scaling, cropping, and normalization operations so that the image size and pixel value meet the requirements of subsequent processing;

[0080] 3) For audio data, perform audio noise reduction and segmentation processing to improve the quality and processability of audio data;

[0081] S13, sending the preprocessed data to the vector conversion module;

[0082] Among them, data sources include relational databases (such as MySQL, Oracle), non-relational databases (such as MongoDB), file systems (such as local files or distributed file systems) and network data sources (such as web pages).

[0083] ②The vector conversion module converts different types of data into vector representations, including:

[0084] 1) For text data, use word embedding models (such as Word2Vec and GloVe) to convert words into corresponding vectors, and then use document vector generation algorithms (such as Doc2Vec) to combine the vectors corresponding to the words into a text feature vector representing the entire document;

[0085] 2) For image data, the image data is input into the convolutional neural network CNN (based on the deep learning framework TensorFlow or PyTorch). The convolutional neural network CNN is initialized with pre-trained weights (such as the weights of VGGNet, ResNet and other models). After multiple layers of convolution and pooling operations, the output of the last fully connected layer is extracted as the image feature vector;

[0086] 3) For audio data, the audio data is input into the recurrent neural network RNN ​​to obtain the audio feature vector, and an audio processing library (such as librosa) is used to assist in the processing of the audio data.

[0087] ③ The Elasticsearch storage and indexing module stores the vectors in the Elasticsearch database and optimizes the indexing algorithm based on the characteristics of the vectors to improve the storage efficiency and retrieval speed of the vectors. It also supports dynamic index updates to ensure that the system can accurately and quickly respond to query requests, including:

[0088] S31. Install and configure the Elasticsearch database and create an index structure specifically for storing vectors:

[0089] For vectors, define the appropriate field type (such as dense_vector) and set the index structure parameters according to the vector dimension and similarity function;

[0090] S32, importing the vectors sent by the vector conversion module into the Elasticsearch database in batches through the API interface of Elasticsearch, and adopting an optimized storage strategy (such as adopting a compression algorithm to reduce storage space occupation) during the import process;

[0091] S33. Regularly optimize and update the index structure to cope with data changes and growth, maintain the stability of retrieval speed, and ensure that the system can respond to query requests accurately and quickly.

[0092] ④ The vector reasoning module implements a variety of vector reasoning operations based on the vector computing function of Elasticsearch to mine the potential relationships between vectors and discover potential patterns and anomalies of vectors, including:

[0093] 1) Use clustering algorithms (such as K-Means clustering algorithm and DBSCAN clustering algorithm) to group vectors. By analyzing the characteristics and distribution of vectors, similar vectors are classified into the same category to explore the potential relationship between vectors and discover the potential patterns of vectors (for example, in the market analysis scenario, clustering consumer behavior vectors can identify the characteristics and preferences of different consumer groups);

[0094] 2) By setting thresholds or using anomaly detection algorithms (such as the isolation forest algorithm) to monitor vectors in real time, anomaly vectors that differ greatly from normal vector patterns can be identified (for example, in the field of network security, anomaly detection of network traffic vectors can promptly detect potential network attack behaviors).

[0095] In the technical solution of the present application, the vector reasoning module monitors data changes and user behavior in real time, dynamically adjusts the vector model and index structure, and automatically triggers the vector conversion module to update the vector when new data is added or the user's query behavior changes, and the Elasticsearch storage and indexing module to update the index structure. At the same time, according to real-time data and user feedback, the parameters of the vector reasoning algorithm are dynamically adjusted to improve the real-time and accuracy of retrieval and analysis (for example, in an e-commerce recommendation system, the user vector and product vector are updated in time according to the user's real-time purchasing behavior and browsing history to optimize the recommendation results).

[0096] ⑤ The user interaction module provides users with a friendly and convenient front-end interactive interface, supports users to input query requests in a variety of ways, calls corresponding search functions according to query requests, and displays information retrieval results in an intuitive way, including:

[0097] S51. Use a front-end development framework (such as Vue.js, React.js) to build a front-end interactive interface, and the user inputs a query request in the query input box by entering text, uploading an image, and uploading an audio;

[0098] S52, the front-end interactive interface sends the query request (via AJAX or other asynchronous communication technologies) to the back-end server, and the back-end server calls the vector reasoning module to perform data retrieval and returns the information retrieval result to the front-end interactive interface;

[0099] S53, after receiving the information retrieval results, the front-end interactive interface presents the information retrieval results to the user in a corresponding form (such as a list, chart, image) according to the data type and result characteristics (for example, in the display of text query results, the titles and abstracts of related documents can be displayed in a list form; for image search results, the images are displayed in a thumbnail form);

[0100] Among them, the front-end interactive interface includes a query input box, a query button and a result display area.

[0101] Specifically, after the vector reasoning module is called by the backend server, it performs data retrieval, including:

[0102] S521, performing semantic analysis on the query request, using the knowledge graph and word vector model to expand the semantic scope of the query keyword, and generating a semantic vector;

[0103] S522, fusing the semantic vector with the original vector corresponding to the query request to generate a query vector;

[0104] S523. Based on the vector calculation function and similarity function provided by Elasticsearch, the similarity between the query vector and the vectors stored in the Elasticsearch database is calculated, sorted according to the similarity, and the most similar result is returned to the backend server.

[0105] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent information retrieval and analysis system based on ES support vector, characterized by: It includes data collection module, vector conversion module, Elasticsearch storage and indexing module, vector reasoning module and user interaction module; The data acquisition module collects data from various data sources, pre-processes the collected data to ensure the quality and consistency of the data, and sends the pre-processed data to the vector conversion module; The vector conversion module converts different types of data into vector representations. By formulating adaptive vector conversion methods for different types of data, it ensures that the vectors can accurately reflect the data characteristics and sends the vectors to the Elasticsearch storage and indexing module. The Elasticsearch storage and indexing module stores vectors in the Elasticsearch database and optimizes the indexing algorithm based on the characteristics of vectors to improve the storage efficiency and retrieval speed of vectors. It also supports dynamic index updates to ensure that the system can respond to query requests accurately and quickly. The vector reasoning module implements various vector reasoning operations based on the vector computing function of Elasticsearch to mine the potential relationships between vectors, discover potential patterns and anomalies of vectors, and provide users with valuable information analysis results. The user interaction module provides users with a friendly and convenient front-end interactive interface, supports users to input query requests in a variety of ways, calls corresponding retrieval functions according to query requests, and displays information retrieval results in an intuitive way.

2. The intelligent information retrieval and analysis system based on support vector of ES according to claim 1 is characterized by: The data acquisition module collects data from various data sources, pre-processes the collected data to ensure the quality and consistency of the data, and sends the pre-processed data to the vector conversion module, including: S11. According to the connection protocols and interfaces of different data sources, corresponding connection parameters are set to establish a stable data collection channel for data collection; S12. Perform corresponding preprocessing operations based on the characteristics of different types of data: 1) For text data, remove stop words and punctuation marks, perform stem extraction, part-of-speech tagging and word segmentation; 2) For image data, perform image scaling, cropping, and normalization operations so that the image size and pixel value meet the requirements of subsequent processing; 3) For audio data, perform audio noise reduction and segmentation processing to improve the quality and processability of audio data; S13, sending the preprocessed data to the vector conversion module; Among them, data sources include relational databases, non-relational databases, file systems and network data sources.

3. The intelligent information retrieval and analysis system based on ES support vector according to claim 1 is characterized in that: The vector conversion module converts different types of data into vector representations, including: 1) For text data, use the word embedding model to convert words into corresponding vectors, and then use the document vector generation algorithm to combine the vectors corresponding to the words into a text feature vector representing the entire document; 2) For image data, the image data is input into the convolutional neural network CNN, which is initialized with pre-trained weights. After multiple layers of convolution and pooling operations, the output of the last fully connected layer is extracted as the image feature vector; 3) For audio data, the audio data is input into the recurrent neural network RNN ​​to obtain the audio feature vector, and the audio processing library is used to assist in the processing of the audio data.

4. The intelligent information retrieval and analysis system based on ES support vector according to claim 1 is characterized in that: The Elasticsearch storage and indexing module stores vectors in the Elasticsearch database, optimizes the indexing algorithm based on the characteristics of the vectors to improve the storage efficiency and retrieval speed of the vectors, and supports dynamic index updates to ensure that the system can accurately and quickly respond to query requests, including: S31. Install and configure the Elasticsearch database and create an index structure specifically for storing vectors: For vectors, define the appropriate field type and set the index structure parameters based on the vector dimension and similarity function; S32, importing the vectors sent by the vector conversion module into the Elasticsearch database in batches through the API interface of Elasticsearch, and adopting an optimized storage strategy during the import process; S33. Regularly optimize and update the index structure to cope with data changes and growth, maintain the stability of retrieval speed, and ensure that the system can respond to query requests accurately and quickly.

5. The intelligent information retrieval and analysis system based on support vector of ES according to claim 1 is characterized by: The vector reasoning module implements various vector reasoning operations based on the vector computing function of Elasticsearch to mine potential relationships between vectors and discover potential patterns and anomalies of vectors, including: 1) Use clustering algorithms to group vectors. By analyzing the characteristics and distribution of vectors, similar vectors are classified into the same category to explore the potential relationships between vectors and discover the potential patterns of vectors. 2) By setting a threshold or using anomaly detection algorithms to monitor vectors in real time, abnormal vectors that differ greatly from normal vector patterns can be identified.

6. The intelligent information retrieval and analysis system based on support vector of ES according to claim 5 is characterized by: The vector reasoning module monitors data changes and user behaviors in real time, dynamically adjusts the vector model and index structure, and automatically triggers the vector conversion module to update the vectors and the Elasticsearch storage and indexing module to update the index structure when new data is added or the user's query behavior changes. At the same time, the parameters of the vector reasoning algorithm are dynamically adjusted according to real-time data and user feedback to improve the real-time and accuracy of retrieval and analysis.

7. The intelligent information retrieval and analysis system based on ES support vector according to claim 1 is characterized in that: The user interaction module provides users with a friendly and convenient front-end interactive interface, supports users to input query requests in a variety of ways, calls corresponding search functions according to the query requests, and displays information retrieval results in an intuitive way, including: S51, using a front-end development framework to build a front-end interactive interface, and the user inputs a query request in the query input box by entering text, uploading an image, and uploading an audio; S52, the front-end interactive interface sends the query request to the back-end server, and the back-end server calls the vector reasoning module to perform data retrieval and returns the information retrieval result to the front-end interactive interface; S53, after receiving the information retrieval result, the front-end interactive interface presents the information retrieval result to the user in a corresponding form according to the data type and result characteristics; Among them, the front-end interactive interface includes a query input box, a query button and a result display area.

8. The intelligent information retrieval and analysis system based on support vector of ES according to claim 7 is characterized by: After being called by the backend server, the vector reasoning module performs data retrieval, including: S521, performing semantic analysis on the query request, using the knowledge graph and word vector model to expand the semantic scope of the query keyword, and generating a semantic vector; S522, fusing the semantic vector with the original vector corresponding to the query request to generate a query vector; S523. Based on the vector calculation function and similarity function provided by Elasticsearch, the similarity between the query vector and the vectors stored in the Elasticsearch database is calculated, sorted according to the similarity, and the most similar result is returned to the backend server.

Images