Information technology retrieval and consultation system and method

Abstract

The present invention relates to the field of data retrieval technology, and in particular to an information technology retrieval consulting system and method. The method comprises the following steps: obtaining user input data; extracting modal key information of the user input data to obtain multimodal key semantic information data; performing data preprocessing on the multimodal key semantic information data to generate standard user input data; performing historical retrieval information retrieval on the standard user input data, and when historical retrieval records are retrieved, performing precise portrait construction on the standard user input data to generate a precise retrieval portrait of the user; when no historical records are retrieved, performing retrieval reasoning on the standard user input data to generate a fuzzy retrieval portrait of the user. The present invention solves the problems of low retrieval accuracy, incomplete information coverage and poor user experience in the prior art by combining multimodal information extraction, precise and fuzzy portrait construction, cross-context inheritance, dynamic knowledge fusion and sentiment analysis optimization.

Description

Technical Field

[0001] The present invention relates to the field of data retrieval technology, and in particular to an information technology retrieval consulting system and method. Background Art

[0002] Initially, information retrieval relied primarily on manual operations and simple keyword matching. Human searchers used indexing and classification to help users find the information they needed. These retrieval methods were simple, relying on manual knowledge and experience, resulting in low efficiency and poor accuracy. With the rise of computer technology and the internet, information retrieval methods underwent fundamental changes. The emergence of search engine technology, such as early AltaVista and Google, enabled automated information retrieval by crawling and indexing web content based on complex algorithmic models. Users could obtain faster and more accurate search results with simple keyword input. With the development of big data and artificial intelligence, information retrieval has expanded beyond static web page searches to multi-dimensional information analysis. Technologies based on natural language processing (NLP), machine learning, and deep learning have been gradually applied to information retrieval systems, enhancing the intelligence level of information retrieval. The application of technologies such as personalized recommendations, semantic search, and intelligent question-answering has made search and inquiry more precise and efficient. However, traditional retrieval methods often process each query independently, failing to effectively inherit previous contextual information. This results in a lack of relevance in successive queries. Furthermore, knowledge retrieval relies on a single data source or static information, which limits retrieval results. Summary of the Invention

[0003] Based on this, it is necessary to provide an information technology retrieval and consultation system and method to solve at least one of the above technical problems.

[0004] To achieve the above-mentioned purpose, an information technology search and consultation method is provided, the method comprising the following steps:

[0005] Step S1: obtaining user input data; extracting modal key information of the user input data to obtain multimodal key semantic information data; performing data preprocessing on the multimodal key semantic information data to generate standard user input data;

[0006] Step S2: Perform historical search information retrieval on the standard user input data. When historical search records are found, a precise profile is constructed for the standard user input data to generate a precise user search profile. When no historical records are found, search reasoning is performed on the standard user input data to generate a fuzzy user search profile. A user input context graph is constructed, and cross-query context inheritance is performed on the user input context graph based on the precise user search profile and the fuzzy user search profile to generate similarity search inheritance data.

[0007] Step S3: Acquire external information data; perform dynamic knowledge retrieval on user input data based on similarity retrieval inheritance data to generate dynamic knowledge retrieval data; perform multi-knowledge source fusion on the dynamic knowledge retrieval data and cross-domain similarity retrieval data to generate fused retrieval data; perform retrieval display adjustment on the fused retrieval data using a preset sentiment analysis model to generate final hybrid retrieval display data;

[0008] Step S4: Perform hybrid search feedback optimization on the final hybrid search display data to perform information technology search consulting optimization operations.

[0009] The present invention extracts multimodal key information (such as text, voice, image, etc.) input by the user and performs data preprocessing, which can unify and standardize user input in different forms, ensure the consistency and accuracy of subsequent processing, and thus lay a solid foundation for the entire retrieval process. By retrieving historical records or generating precise or fuzzy retrieval portraits of users through reasoning, user needs can be described in a personalized way and retrieval accuracy can be improved. Especially when there are no historical records, fuzzy retrieval portraits can be generated through reasoning to ensure the system's processing capabilities for new users or small amounts of data. At the same time, the inheritance mechanism across query contexts can continuously optimize the contextual relevance of user input and improve the relevance and consistency of continuous queries. Through dynamic knowledge retrieval and cross-domain information fusion, broader knowledge coverage can be achieved between multiple information sources, avoiding the limitations caused by a single data source or static information. This fusion not only improves the comprehensiveness and accuracy of the retrieval, but also allows for more flexible integration of data from different fields, ensuring higher quality retrieval results. In addition, the sentiment analysis model adjusts the display method according to the user's emotional tendencies, thereby improving user experience and satisfaction. Through continuous hybrid search feedback optimization, the search strategy can be adjusted based on real-time user feedback, dynamically optimizing the information technology search process. This feedback mechanism ensures that the search system continuously improves during operation, thereby enhancing the system's intelligence level and personalized service capabilities. Therefore, by combining multimodal information extraction, precise and fuzzy portrait construction, cross-context inheritance, dynamic knowledge fusion, and sentiment analysis optimization, the present invention solves the problems of low search accuracy, incomplete information coverage, and poor user experience in existing technologies.

[0010] Preferably, step S1 includes the following steps:

[0011] Step S11: obtaining user input data, wherein the user input data includes text input data, image input data and voice input data;

[0012] Step S12: extracting modal key information from the text input data, the image input data, and the voice input data to obtain multimodal key information extraction data;

[0013] Step S13: performing unified semantic representation on the multimodal key information extraction data to generate multimodal key semantic information data;

[0014] Step S14: preprocess the multimodal key semantic information data to generate standard user input data, wherein the data preprocessing includes data missing value prompts, data abnormal format conversion, and data duplicate value removal.

[0015] The present invention can fully understand the user's intentions and needs by acquiring text, image and voice data and extracting key information. The extraction of this multimodal information enables the system to understand text, visual and voice content at the same time, providing a more accurate response. Converting data of different modalities into a unified semantic representation can reduce ambiguity or information missing problems caused by different data sources. The unified semantic representation helps to improve the efficiency and accuracy of the system when processing multiple data sources. By handling missing data, abnormal format conversion and removing duplicate values, the quality of input data is ensured. This process can avoid system errors caused by data problems and improve the stability and reliability of the system. Through efficient multimodal data processing, the user's input, regardless of its form, can be quickly and accurately understood and responded to by the system, thereby improving interaction efficiency and user experience.

[0016] Preferably, step S2 includes the following steps:

[0017] Step S21: Perform historical search information retrieval on the standard user input data. When historical search records are found, retrieve the user's historical search data, and construct a user profile based on the user's historical search data to generate an accurate user search profile.

[0018] Step S22: When no existing historical records are retrieved, user intent recognition is performed on the standard user input data to obtain user intent recognition data; the user intent recognition data is used to perform retrieval reasoning on the standard user input data to generate a user fuzzy retrieval profile;

[0019] Step S23: constructing a context graph for the standard user input data to obtain a user input context graph; dynamically updating the graph context complexity weight of the user input context graph based on the user precise retrieval profile and the user fuzzy retrieval profile to generate graph context dynamic update data;

[0020] Step S24: Calculate the correlation edge nodes of the graph context dynamic update data, and perform cross-query context inheritance on the user input context graph through the correlation edge nodes to generate similarity retrieval inheritance data.

[0021] By retrieving a user's historical data, the present invention can effectively construct a precise user profile. This history-based personalized recommendation system helps provide more customized services and enhance the user experience. For example, when a user's input is relevant to historical search content, the system can quickly provide information related to the user's historical behavior, thereby improving retrieval efficiency and accuracy. In the absence of historical records, a fuzzy retrieval profile of the user can be generated by identifying user intent and performing retrieval inference. Through this process, even without historical data, the system can still identify the user's current needs and construct a preliminary user profile, helping the system understand and predict the user's potential needs. By constructing a context graph of user input and combining the user's precise and fuzzy profiles, the complexity weight of the context graph is dynamically updated. This update not only improves the graph's accuracy and comprehensiveness, but also allows real-time adjustments based on changes in user behavior and needs, enabling the system to respond more intelligently to users. By dynamically updating the relevant edge nodes of the graph context data, context inheritance across queries can be achieved. This mechanism establishes connections between different queries, allowing the system to understand and leverage previous query information to provide more consistent and continuous services, further improving retrieval efficiency and intelligence.

[0022] Preferably, step S22 includes the following steps:

[0023] Step S221: When no existing historical records are retrieved, keyword part-of-speech extraction is performed on the standard user input data to obtain keyword part-of-speech extraction data; sentiment classification is performed on the keyword part-of-speech extraction data to generate keyword sentiment classification data;

[0024] Step S222: performing a tendency analysis on the standard user input data using the keyword sentiment classification data to generate user intention recognition data; performing intent classification on the user intention recognition data to obtain user intention classification data;

[0025] Step S223: performing behavior matching based on the user intention classification data to obtain user behavior data; performing user preference association inference based on the user behavior data to obtain user preference inference data;

[0026] Step S224: Cluster users based on user preference inference data to obtain user fuzzy retrieval portraits.

[0027] The present invention can help the system more accurately understand the user's emotional tendencies and context by extracting keyword parts of speech and performing sentiment classification on standard user input data in step S221. For example, sentiment classification helps distinguish whether the user is in a positive, negative, or neutral mood, which is crucial for providing appropriate feedback or services. The system can adjust the tone and content of the response based on this emotional information, making the interaction more personalized and humane. By performing a tendency analysis on the keyword sentiment classification data and further generating user intent recognition data, this process helps to accurately identify the user's core needs and objectives and provide responses that better meet the user's expectations. Through intent recognition, the system can avoid producing results that do not meet the user's actual needs and improve the relevance and quality of services. Behavior matching is performed based on the user intent classification data to derive user behavior patterns. In addition, through preference association reasoning, the system can further understand the user's preferences and interests. This step helps to enhance the system's recognition of the user's personalized needs, allowing the system to accurately predict the user's future needs and provide more appropriate suggestions or content on this basis. By clustering user preference inference data, users can be divided into different groups based on the similarities of their behaviors and needs, generating fuzzy search profiles of users. This process helps build more detailed user profiles, enabling the system to provide customized search results based on these profiles, improving the user experience. This is especially true when precise historical records are lacking, allowing services with high accuracy to be provided. By integrating analysis across multiple steps, step S22 provides the system with deeper user insights. The system can better understand users' emotions, intentions, behaviors, and preferences, automatically generating the most relevant search profiles when users enter information, and optimizing the accuracy of recommendations and searches.

[0028] Preferably, step S24 includes the following steps:

[0029] Step S241: Performing initial node screening on the user input context graph based on the graph context dynamic update data to obtain initial screening nodes; performing boundary point relevance scoring on the initial screening nodes using a graph algorithm to obtain a relevance score for each initial screening node;

[0030] Step S242: Filter out the edge nodes with the highest correlation according to the correlation score of each initial screening node, and mark them as correlation edge nodes; filter the edge-connected nodes of the initial screening nodes through the correlation edge nodes to obtain adjacent associated screening nodes;

[0031] Step S243: performing node directionality analysis on the adjacent associated screening nodes to generate adjacent associated screening node directionality data; performing context node marking on the adjacent associated screening nodes based on the adjacent associated screening node directionality data to obtain previous associated nodes and following associated nodes;

[0032] Step S244: Calculate the node information entropy value of the correlation edge node through the previous related node and the following related node to obtain the standard node information entropy value; perform cross-query context inheritance on the user input context graph through the standard node information entropy value to generate similarity retrieval inheritance data.

[0033] The present invention performs initial node screening based on the dynamic update data of the graph context through step S241, and scores the relevance of the nodes, which can effectively identify the nodes most relevant to the user input. This screening and scoring mechanism based on the graph algorithm helps the system focus on the nodes with the most information value, thereby improving the efficiency and accuracy of data processing. Step S242 screens out the most relevant edge nodes based on the relevance score of each initial screening node, and performs edge connection node screening. This process can establish a more accurate node connection relationship by identifying key correlation edge nodes, optimize the structure of the graph, and improve the accuracy of information retrieval. Step S243 generates directional data of adjacent associated screening nodes through node directional analysis, and performs context node marking. This analysis ensures that the system can deeply understand the contextual relationship between nodes, thereby distinguishing the associated information between the previous and following contexts. By marking these context nodes, the system can more flexibly adjust the extraction and use of information and enhance the understanding of user needs. Step S244 implements cross-query context inheritance by calculating the information entropy value of the standard node. This process quantifies the uncertainty of node information and dynamically updates the graph based on this information entropy value, allowing similar queries to share and inherit relevant contextual data, thereby improving the efficiency and accuracy of similarity retrieval. Through the above steps, the system can dynamically adjust and optimize the retrieval process based on the contextual information entered by the user. In particular, when processing complex or changing queries, it can more intelligently perform similarity inheritance and information optimization, enabling the system to provide more efficient, accurate, and personalized retrieval results, improving the user experience.

[0034] Preferably, performing cross-query context inheritance on the user input context graph by using node information entropy values ​​includes:

[0035] Perform context topology analysis on the user input context graph to obtain context structure topology data; perform semantic inheritance relationship extraction on the context structure topology data to generate semantic inheritance relationship data;

[0036] Perform query pattern matching on semantic inheritance relationship data through node information entropy value to generate query pattern matching data; perform propagation path analysis on user input context graph based on query pattern matching data to generate context query propagation path data;

[0037] The standard node information entropy value is used to verify the cross-node inheritance rules of the context query propagation path data. When the information entropy value of the node in the context query propagation path is less than or equal to the standard node information entropy value, similarity retrieval inheritance data is generated; otherwise, the corresponding context query propagation path is eliminated.

[0038] The present invention obtains context structure topology data by performing a topological structure analysis on the context graph input by the user. Through this analysis, the system can identify the relationship and dependency between nodes, laying the foundation for the subsequent extraction of semantic inheritance relationships. The generation of semantic inheritance relationship data ensures that context information can be effectively transmitted and continued during the query process, thereby providing the system with a more accurate context understanding. Matching the semantic inheritance relationship data with the query pattern through the node information entropy value helps the system identify the query intention input by the user, and builds a matching pattern on this basis. This matching mechanism enables the system to quickly determine the relationship between the user's current query intention and the historical query pattern, providing an accurate basis for subsequent context propagation path analysis. By generating context query propagation path data, the system can better handle the information flow and association between multiple queries, thereby improving retrieval efficiency. The standard node information entropy value is used to verify the node information entropy value in the context query propagation path, thereby determining whether to execute the cross-node inheritance rule. If a node's information entropy value is lower than the standard node's information entropy value, it is considered to meet the inheritance rules and generate similarity search inheritance data. Otherwise, the system removes the node to prevent irrelevant or low-relevance nodes from affecting query results. This dynamic verification mechanism optimizes the accuracy of information inheritance, ensuring that only relevant and high-value node information is transferred and inherited. Through cross-query context inheritance, the system effectively maintains and extends contextual consistency within queries. In particular, when multiple queries have similar information, the system automatically inherits existing contextual relationships to avoid duplicate calculations or information loss. This mechanism improves the accuracy of system responses, especially in complex queries or multi-turn conversations, allowing it to accurately understand user intent and provide more precise results. By dynamically adjusting node information entropy values, the system can flexibly adapt to different types of queries, ensuring that information in the query path is effectively propagated while irrelevant nodes or erroneous information are removed. This enhances the system's adaptability to complex queries and diverse scenarios, enabling it to more intelligently generate similarity search results based on the context.

[0039] Preferably, step S3 includes the following steps:

[0040] Step S31: using an external domain knowledge base to obtain external information data;

[0041] Step S32: performing a cross-domain similarity search on the user input data according to the similarity search inheritance data to obtain cross-domain similarity search data; performing a search field analysis on the cross-domain similarity search data to generate search field analysis data;

[0042] Step S33: performing dynamic knowledge retrieval on the cross-domain similar retrieval data based on the retrieval domain analysis data to generate dynamic knowledge retrieval data; performing multi-knowledge source fusion on the dynamic knowledge retrieval data and the cross-domain similar retrieval data to generate fused retrieval data;

[0043] Step S34: Perform user input context sentiment analysis on the standard user input data through a preset sentiment analysis model to generate user input context sentiment data; use the user input context sentiment data to perform retrieval display adjustment on the fused retrieval data to generate the final hybrid retrieval display data.

[0044] The present invention uses an external domain knowledge base to obtain external information data through step S31. The system can quickly integrate external expertise and improve the depth and breadth of information processing. This integration not only enables the system to better understand and process the data input by the user, but also provides more comprehensive background information for subsequent cross-domain similarity retrieval. Step S32 performs a cross-domain similarity retrieval on the user input data by similarity retrieval inheritance data, ensuring that the user input information can be compared and analyzed across different domains. This can find information related to the user query but existing in other domains, broadening the retrieval perspective. In addition, the generation of retrieval domain analysis data helps the system understand the diversity of information sources and domain characteristics, thereby more accurately performing retrieval and processing. Step S33 performs dynamic knowledge retrieval based on the retrieval domain analysis data and generates fused retrieval data through multi-knowledge source fusion. This step not only responds to changes in user input in real time, but also provides more accurate answers based on the fusion of knowledge from different domains. By fusing different knowledge sources, the system can obtain more comprehensive and accurate information, avoiding the limitations of a single source and improving the multidimensionality and accuracy of the answers. Step S34 performs sentiment analysis on the context of the user input through the sentiment analysis model, which can capture the emotional tendency of the user when expressing. The fused retrieval data is displayed and adjusted according to the sentiment analysis results to ensure that the final retrieval display data is not only accurate but also meets the emotional needs of the user. This emotion-based display optimization can improve the user's interactive experience, especially in application scenarios that require emotional resonance, which increases the intelligence and humanity of the system. Through the above steps, the system can cross domains, integrate multiple knowledge sources, and provide richer retrieval results. At the same time, the addition of sentiment analysis makes the displayed data more in line with the user's emotional tendencies, thereby improving the overall user experience. This method ensures the accuracy, breadth and personalization of information retrieval, allowing users to quickly get answers that meet their needs, while also being more in line with the context and emotional needs of the situation.

[0045] Preferably, performing user input context sentiment analysis on standard user input data using a preset sentiment analysis model includes:

[0046] Performing sentiment dictionary matching on the standard user input data to generate sentiment dictionary matching data; performing sentiment vocabulary extraction on the standard user input data based on the sentiment dictionary matching data to obtain sentiment vocabulary data;

[0047] Perform sentiment polarity analysis based on sentiment vocabulary data and determine the sentiment tendency in standard user input data to obtain sentiment polarity data; perform contextual sentiment enhancement on the sentiment polarity data using a preset sentiment analysis model to generate contextual sentiment enhancement data;

[0048] Contextual emotion enhancement data is used to evaluate the emotion intensity of standard user input data and generate user input context emotion data.

[0049] Through sentiment lexicon matching and sentiment vocabulary extraction, the present invention enables the system to efficiently capture emotional information in text, reducing the risk of misjudgment. Sentiment lexicon matching ensures that emotion-related words are not missed during the analysis process, thereby improving the accuracy of sentiment analysis. Contextual sentiment enhancement, by combining a preset model with the specific context of user input, enables a more accurate understanding and interpretation of emotional tendencies. Compared to simple sentiment vocabulary analysis, context-based sentiment enhancement can effectively identify implicit emotions (e.g., emotional expressions such as sarcasm and irony), thereby improving the ability to handle complex contexts. Through sentiment intensity assessment, the system can not only determine whether the user's input sentiment is positive, negative, or neutral, but also further assess the intensity of the emotion, which helps the system provide more accurate and personalized responses based on the intensity of the user's emotions. The system can adjust its response based on the user's emotional data. For example, when the user's emotions are detected to be relatively negative, the system can adjust to more soothing or positively motivating language, providing a more humane interactive experience. This emotional intelligence enhancement helps improve user satisfaction with the system and the quality of interaction. The results of sentiment analysis can be used to optimize recommendation systems, advertising, user services, and other scenarios. For example, in customer support or social media monitoring, the system can make adjustments based on user sentiment feedback, improve service strategies, and increase user loyalty and satisfaction.

[0050] Preferably, step S4 includes the following steps:

[0051] Step S41: collecting user consultation feedback on the final hybrid search display data to obtain user consultation feedback data, wherein the user consultation feedback collection includes click feedback collection and dwell time feedback collection;

[0052] Step S42: performing hybrid search optimization on the final hybrid search display data based on the user consultation feedback data to perform information technology search consultation optimization operations.

[0053] The present invention collects user consultation feedback on the final hybrid search display data in step S41. This allows the system to fully understand how users interact with the display data, specifically including click feedback and dwell time feedback. This feedback provides valuable user behavior data, helping the system analyze user interests and areas of focus, thereby enabling more accurate adjustment and optimization of the search display content. Click feedback reflects user interest in certain content, while dwell time provides information on user engagement with the content and the degree of in-depth reading. Step S42 optimizes the final hybrid search display data based on the collected user consultation feedback data. This optimization process can adjust the relevance and display format of the search results, allowing the system to better adapt to user needs and preferences. For example, the system can adjust rankings based on click volume or optimize the order and format of content display based on dwell time, thereby improving the accuracy and appeal of the search results. Through optimization based on user feedback, the system can gradually improve the search algorithm and display strategy, providing a search experience that better meets user needs. The long-term feedback loop can help the system continuously adjust and improve the relevance and personalization of the search, thereby effectively improving user satisfaction and user stickiness of the system. By optimizing the feedback from the final displayed data, the system can continuously adjust its information technology retrieval strategy and improve its retrieval algorithms, making the system more efficient, accurate, and intelligent when processing information queries. This not only improves retrieval efficiency but also provides users with more customized consulting optimization based on their feedback, further enhancing the intelligence level of the information retrieval system. This optimization process enables the system to dynamically adapt to the needs and behavioral changes of different users. By continuously monitoring and collecting user behavior data, the system can more flexibly adjust its display strategy and respond to user needs in real time, thereby maintaining efficient and personalized service.

[0054] In this specification, an information technology search and consultation system is provided for executing the above-mentioned information technology search and consultation method. The information technology search and consultation system includes:

[0055] The data processing module is used to obtain user input data; extract the modal key information of the user input data to obtain multimodal key semantic information data; perform data preprocessing on the multimodal key semantic information data to generate standard user input data;

[0056] The retrieval inheritance module is used to retrieve historical retrieval information of standard user input data. When historical retrieval records are retrieved, a precise portrait of the standard user input data is constructed to generate a precise retrieval portrait of the user. When no historical records are retrieved, retrieval reasoning is performed on the standard user input data to generate a fuzzy retrieval portrait of the user. A user input context map is constructed, and cross-query context inheritance is performed on the user input context map based on the precise retrieval portrait and the fuzzy retrieval portrait to generate similarity retrieval inheritance data.

[0057] The fusion retrieval module is used to obtain external information data; perform dynamic knowledge retrieval on user input data based on similarity retrieval inheritance data to generate dynamic knowledge retrieval data; perform multi-knowledge source fusion on dynamic knowledge retrieval data and cross-domain similarity retrieval data to generate fusion retrieval data; perform retrieval display adjustment on the fusion retrieval data through a preset sentiment analysis model to generate the final hybrid retrieval display data;

[0058] The retrieval feedback module is used to perform hybrid retrieval feedback optimization on the final hybrid retrieval display data to perform information technology retrieval consulting optimization operations.

[0059] The beneficial effect of the present invention is that by obtaining user input and extracting multimodal key information, different forms of user input (such as text, images, voice, etc.) are converted into standardized data formats to ensure the consistency and processability of the input data. This preprocessing process helps the subsequent retrieval system to understand user needs more accurately and provide a reliable data foundation for accurate retrieval and reasoning. By generating accurate or fuzzy retrieval portraits of users through historical retrieval records and reasoning, personalized demand analysis is achieved. When there are no historical records, fuzzy portraits are generated through reasoning to ensure that the system can handle new users or small amounts of data. At the same time, through cross-query context inheritance technology, the relevance in continuous queries is improved, the context consistency during multiple interactions is guaranteed, and the accuracy of retrieval is further enhanced by similarity retrieval inheritance data. Through dynamic knowledge retrieval and cross-domain information fusion, relevant information can be obtained from multiple knowledge sources, improving the breadth and depth of retrieval. Multi-knowledge source fusion avoids the problem of information islands and increases the system's coverage of multi-domain data. At the same time, the introduction of the sentiment analysis model enables the retrieval results to be adjusted according to the user's emotional tendencies, thereby improving the personalization of the retrieval results and user experience. Through retrieval feedback optimization, the module can dynamically adjust the retrieval strategy and results according to user feedback, thereby improving the intelligence and adaptability of the retrieval system. The continuous optimization process ensures that the system is constantly improved, improves the relevance, accuracy and real-time performance of query results, and enhances the learning ability of the system so that future retrieval can better meet user needs. Therefore, the present invention solves the problems of low retrieval accuracy, incomplete information coverage and poor user experience in the prior art by combining multimodal information extraction, precise and fuzzy portrait construction, cross-context inheritance, dynamic knowledge fusion and sentiment analysis optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 A flowchart of the steps of an information technology search and consultation method;

[0061] Figure 2 for Figure 1 Detailed implementation steps of step S2 in FIG.

[0062] Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG.

[0063] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0064] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. It is obvious that the embodiments described are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present invention.

[0065] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0066] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0067] To achieve this, please refer to Figures 1 to 3 , an information technology search and consultation method, the method comprising the following steps:

[0068] Step S1: obtaining user input data; extracting modal key information of the user input data to obtain multimodal key semantic information data; performing data preprocessing on the multimodal key semantic information data to generate standard user input data;

[0069] Step S2: Perform historical search information retrieval on the standard user input data. When historical search records are found, a precise profile is constructed for the standard user input data to generate a precise user search profile. When no historical records are found, search reasoning is performed on the standard user input data to generate a fuzzy user search profile. A user input context graph is constructed, and cross-query context inheritance is performed on the user input context graph based on the precise user search profile and the fuzzy user search profile to generate similarity search inheritance data.

[0070] Step S3: Acquire external information data; perform dynamic knowledge retrieval on user input data based on similarity retrieval inheritance data to generate dynamic knowledge retrieval data; perform multi-knowledge source fusion on the dynamic knowledge retrieval data and cross-domain similarity retrieval data to generate fused retrieval data; perform retrieval display adjustment on the fused retrieval data using a preset sentiment analysis model to generate final hybrid retrieval display data;

[0071] Step S4: Perform hybrid search feedback optimization on the final hybrid search display data to perform information technology search consulting optimization operations.

[0072] The present invention extracts multimodal key information (such as text, voice, image, etc.) input by the user and performs data preprocessing, which can unify and standardize user input in different forms, ensure the consistency and accuracy of subsequent processing, and thus lay a solid foundation for the entire retrieval process. By retrieving historical records or generating precise or fuzzy retrieval portraits of users through reasoning, user needs can be described in a personalized way and retrieval accuracy can be improved. Especially when there are no historical records, fuzzy retrieval portraits can be generated through reasoning to ensure the system's processing capabilities for new users or small amounts of data. At the same time, the inheritance mechanism across query contexts can continuously optimize the contextual relevance of user input and improve the relevance and consistency of continuous queries. Through dynamic knowledge retrieval and cross-domain information fusion, broader knowledge coverage can be achieved between multiple information sources, avoiding the limitations caused by a single data source or static information. This fusion not only improves the comprehensiveness and accuracy of the retrieval, but also allows for more flexible integration of data from different fields, ensuring higher quality retrieval results. In addition, the sentiment analysis model adjusts the display method according to the user's emotional tendencies, thereby improving user experience and satisfaction. Through continuous hybrid search feedback optimization, the search strategy can be adjusted based on real-time user feedback, dynamically optimizing the information technology search process. This feedback mechanism ensures that the search system continuously improves during operation, thereby enhancing the system's intelligence level and personalized service capabilities. Therefore, by combining multimodal information extraction, precise and fuzzy portrait construction, cross-context inheritance, dynamic knowledge fusion, and sentiment analysis optimization, the present invention solves the problems of low search accuracy, incomplete information coverage, and poor user experience in existing technologies.

[0073] In the embodiment of the present invention, reference Figure 1 FIG. 1 is a flow chart showing the steps of an information technology search and consultation method according to the present invention. In this example, the information technology search and consultation method includes the following steps:

[0074] Step S1: obtaining user input data; extracting modal key information of the user input data to obtain multimodal key semantic information data; performing data preprocessing on the multimodal key semantic information data to generate standard user input data;

[0075] In the embodiments of the present invention, user input data can come from multiple sources and modalities, such as text, speech, images, videos, sensor data, etc. The system first receives the user's input data through different input interfaces (such as text boxes, speech recognition, image uploads, etc.). For each input modality, different data acquisition methods are adopted: collecting the text data input by the user on the interface, converting speech data into text data through speech recognition technology, and obtaining the image or video files provided by the user through image / video acquisition tools. Integrate the user input data of different modalities into a unified processing system. At this time, the data has not been preprocessed or parsed and is usually stored in its original format, such as text, image files, speech audio files, etc. Through natural language processing (NLP) technology, extract the key information in the text, such as keyword extraction, entity recognition, syntactic analysis, sentiment analysis, etc. After converting speech into text using speech recognition technology, further use a semantic understanding model to extract the key information in the speech, such as extracting the theme, sentiment, etc. Use image processing technology and computer vision algorithms (such as convolutional neural networks, CNNs) to extract key features or objects from images. For example, image recognition, object detection, scene analysis, etc.; video input involves image processing for each frame and analysis of the inter-frame relationship. Integrate the key information extracted from different modalities. Use multimodal learning technologies (such as multimodal neural networks, cross-modal mapping models, etc.) to merge the information of text, images, speech, and other modalities to form multimodal key semantic information data. For example, if the user input contains both text descriptions and images, the system can combine the visual features of the images with the semantic information of the text to provide a richer context understanding. For text in Chinese or other languages, first perform word segmentation to split the sentence into individual words. Remove common words with no practical meaning, such as "de", "le", etc. Label the词性 information for each word, such as nouns, verbs, adjectives, etc. Convert the text into a vector representation (such as using pre-trained models like Word2Vec, BERT, etc.) for subsequent model processing. Standardize the data of different modalities to ensure that they are within the same range, have consistent dimensions and formats. For example, convert images and text into vectors of the same length through predefined models, and speech and sensor data can be converted into numerical vectors through corresponding feature extraction steps. According to the system requirements, organize the preprocessed multimodal data into standard user input data in a unified format, which can be a unified multi-dimensional array, vector set, database record, etc. For example, the standard input data is a comprehensive data structure containing text vectors, image feature vectors, speech feature vectors, etc., for use by subsequent analysis and decision-making modules.

[0076] It should be noted that there is an unclear part "词性" in the original text which is translated as "词性" here. You may need to correct it to the appropriate Chinese character for accurate translation.Step S2: Perform historical search information retrieval on the standard user input data. When historical search records are found, a precise profile is constructed for the standard user input data to generate a precise user search profile. When no historical records are found, search reasoning is performed on the standard user input data to generate a fuzzy user search profile. A user input context graph is constructed, and cross-query context inheritance is performed on the user input context graph based on the precise user search profile and the fuzzy user search profile to generate similarity search inheritance data.

[0077] In an embodiment of the present invention, a historical search is performed on standard user input data (including user input in modalities such as text, images, and voice). The historical search data includes the user's past behavior data, search history, interaction data, etc. Natural language processing (NLP), speech recognition, image matching, and other technologies are used to match the standard user input data with the historical data. If a historical record is retrieved, that is, the standard user input data matches the historical search data (a matching query record exists), step S2.2 is entered. If no matching historical record is found, that is, the standard user input data has no matching item in the historical record, step S2.3 is entered. Based on the historical search records and user behavior data, the user's search interests, preferences, past search paths, etc. are analyzed. Combined with the user's behavior in the historical record (such as clicks, browsing time, browsing content, etc.), an accurate user profile is created for the user. Based on the historical record, the user's basic attributes, such as age, gender, interest tags, usage habits, etc., are extracted and integrated into the user profile. The user's potential needs, intentions, and purposes are analyzed based on the historical record, and the user profile is enhanced with emotion, preference, or intention features related to their current needs. During the precise profile construction process, a personalized and accurate user search profile is constructed by combining the user's past query history, behavior patterns, and preference data. This profile includes the user's interests and preferences, frequently searched keywords, topics or domains, the user's sentiment toward specific content (e.g., positive or negative sentiment), and relevant contextual information (e.g., time, location, device type, etc.). If the system fails to find a match within historical search records, it performs inference based on the current standard user input data, utilizing inference models (e.g., machine learning or deep learning models) to infer the user's potential needs and interests. By analyzing the contextual information of the user input (e.g., time, topic, sentiment of the current session) and the user's behavior patterns, the system infers the user's needs or query direction. Based on the characteristics of the user's current input data (e.g., semantic similarity of the query, sentiment analysis results, etc.), a fuzzy match is performed within known datasets or external domains to infer the user's interests or intentions. In the absence of historical search records, the inference process generates a fuzzy user profile, referred to as a "user fuzzy search profile." Based on the standard user input data and its historical or inference results, a graph structure is constructed that incorporates various contextual information. Nodes include user-input keywords, topics, sentiment, behavior patterns, etc. In a graph, the edges connecting different nodes represent various relationships, such as semantic relationships (such as synonyms and related words), temporal relationships (such as the time of the user's last query), and contextual relationships (such as the content of similar queries). The graph is optimized to ensure that the relationships between different modal user input data and historical behavior are accurately reflected. For example, data from text, images, and voice input can be fused together using technologies such as embedding vectors and graph neural networks to form a multimodal contextual graph.Leveraging information from the user's precise and fuzzy search profiles, the context graph of the user's input is inherited across queries, resulting in more accurate query results. Inheritance rules are set based on the user's historical query records and inferred profile. For example, if a user has a high interest in a particular field, nodes related to that field are weighted in the context graph and passed on to subsequent queries. During cross-query inheritance, the system updates graph nodes based on the precise and fuzzy profiles. For example, the system may infer a user's ongoing interest in certain topics, thereby increasing the weight of these topics in the current query and promoting the display of relevant content. The similarity between the current user's query and previous or other user queries is calculated by analyzing the similarity between previous queries and the current input, or based on information from the user profile. Based on similarity metrics (such as cosine similarity and Euclidean distance), the user's input data is matched with historical records or other relevant data to generate similarity retrieval inheritance data. Based on this similarity retrieval inheritance data, relevant recommendations, query results, or content display are provided to the user. For example, if the user's query is highly matched with historical data, the system can directly display historically relevant results; if there is no historical data, recommendations are made based on similarity inheritance results.

[0078] Step S3: Acquire external information data; perform dynamic knowledge retrieval on user input data based on similarity retrieval inheritance data to generate dynamic knowledge retrieval data; perform multi-knowledge source fusion on the dynamic knowledge retrieval data and cross-domain similarity retrieval data to generate fused retrieval data; perform retrieval display adjustment on the fused retrieval data using a preset sentiment analysis model to generate final hybrid retrieval display data;

[0079] In an embodiment of the present invention, external information data sources are selected, including open data sets, third-party databases, industry reports, online APIs, etc. These data sources include various types of data such as text, images, audio, and video. External data is obtained through web crawling, API calls, database queries, etc. The obtained data can be structured (such as JSON, XML format) or unstructured (such as web page content, documents, etc.). User input data and similarity search inheritance data are used as search inputs, and dynamic search is performed in combination with context graphs, user profiles, and other information. Through a preset search engine (such as based on inverted indexes, TF-IDF, BERT, and other deep learning models), more accurate searches are performed through semantic search based on user queries and historical search records. User input data is dynamically searched through a cross-domain multi-level knowledge base, and the search results can include literature, reports, expert answers, external API data, etc. in known fields. The retrieved relevant data is organized according to certain sorting rules, involving keyword matching, semantic relevance scoring, etc. Dynamic search results can include text, video links, documents, expert recommendations, and other content. Retrieval results are filtered, sorted, and deduplicated based on factors such as data relevance, user urgency, and preferences, ensuring that the most relevant and valuable information is provided to users. During knowledge retrieval, information from different domains and sources has different forms and structures. First, these knowledge sources are identified and annotated. For example, search results from external databases, recommendations from user behavior analysis, and inference results from historical data can be used. Data from different knowledge sources can be combined using methods such as weighted fusion, priority fusion, and consensus fusion. Specifically, different weights can be assigned to each source based on its reliability, relevance, and user preferences. Information from highly reliable sources (such as expert recommendations and official literature) is prioritized. When multiple sources provide consistent information, its credibility is enhanced and prioritized. Machine learning (such as random forests and XGBoost) or deep learning models (such as neural networks and graph convolutional networks) are used to automatically fuse multiple data sources and output the final fused data. Semantic alignment of data from different knowledge sources ensures consistency within the same context and avoids information clutter. The final search results are generated based on the fused data. These results include relevant data from multiple knowledge sources while maintaining semantic consistency across the data sources. A sentiment lexicon (e.g., a sentiment-based vocabulary or sentiment analysis tool) is used to perform sentiment analysis on the fused search data to identify the sentiment tendencies (e.g., positive, negative, neutral) within the data. Based on the identified sentiment words, a contextual sentiment analysis model is used to contextually modify the sentiment words to improve the accuracy of the sentiment analysis. For example, certain words express different sentiment tendencies in different contexts, necessitating further contextual correction of the sentiment results.The sentiment orientation and intensity of each search result are evaluated to ensure that emotional data that meets user needs is highlighted during display. For example, users prefer positive feedback, so sentiment intensity assessment can be used to increase the weight of positive data. For data with negative sentiment, the system can lower its priority or apply sentiment adjustment during display (such as reminding users of negative content). Based on the sentiment analysis results, the display order and content of the final fused search data are adjusted. Through sentiment filtering, data that meets the user's current emotional needs is presented to the user. For example, positive data is displayed or negative content is filtered based on the user's emotional needs. During the display, the display order of content can be dynamically adjusted, and even filtering can be performed based on sentiment analysis results to enhance the user experience. Based on the fused search results and sentiment analysis adjustments, a clear and concise display structure is designed. Various display formats (such as cards, lists, and grids) can be used to tailor the display to the data type and user preferences. Different display strategies are adopted for data of different modalities (such as text, images, and videos). For example, for text content, a concise summary is used; for image or video content, thumbnails are displayed with clickable details. The final mixed retrieval display data is presented to the user through a preset sorting algorithm (such as sorting adjusted based on sentiment analysis, sorting adjusted based on user portraits, etc.).

[0080] Step S4: Perform hybrid search feedback optimization on the final hybrid search display data to perform information technology search consulting optimization operations.

[0081] In an embodiment of the present invention, the content clicked by users in the displayed data is recorded. Click behavior reflects the user's interest in specific content and its relevance, which can be collected through a click tracking system (such as an event listener). The amount of time a user dwells on a specific displayed item is recorded. Dwell time is a direct reflection of the user's interest in the displayed content, especially for detailed information or long documents. Dwell time data can be collected through user behavior tracking tools in front-end applications. User emotional feedback on specific displayed content is collected through simple sentiment scoring (such as like, dislike, approve, disapprove, etc.) to supplement click and dwell time data. If the system allows, users can also provide more specific opinions through feedback forms, questionnaires, real-time ratings, etc. to help the system understand user needs. User click feedback, dwell time feedback, and sentiment feedback are integrated to generate user consultation feedback data. A feedback score is generated for each displayed item based on different dimensions of the feedback data (such as number of clicks, dwell time, and sentiment orientation), which is used to analyze user interest and satisfaction with the search results. The displayed items clicked by users and their relevance are analyzed to identify which displayed items have a high click-through rate. For these high-click display items, the system can increase their display weight to increase their frequency of appearance in future search results. The dwell time is associated with the relevant content of the display item. Content that is stayed for a long time represents the user's deep attention to the content. The system can mark display items with longer dwell time and increase their display priority. By analyzing emotional feedback (such as the distribution of positive and negative emotions), the user's emotional preferences can be identified. For example, when the user shows obvious negative emotions towards certain types of data, the system can adjust the display frequency of such data accordingly. Through a weighting mechanism, different weights are assigned to click feedback, dwell time feedback, and emotional feedback to ensure that important feedback factors have a greater impact on the optimization process. For example, display items with higher dwell time and click feedback can be given higher weights, thereby strengthening the display of such content in future searches. Dynamic optimization is performed based on feedback data to fine-tune the current mixed retrieval display data.

[0082] Preferably, step S1 includes the following steps:

[0083] Step S11: obtaining user input data, wherein the user input data includes text input data, image input data and voice input data;

[0084] Step S12: extracting modal key information from the text input data, the image input data, and the voice input data to obtain multimodal key information extraction data;

[0085] Step S13: performing unified semantic representation on the multimodal key information extraction data to generate multimodal key semantic information data;

[0086] Step S14: preprocess the multimodal key semantic information data to generate standard user input data, wherein the data preprocessing includes data missing value prompts, data abnormal format conversion, and data duplicate value removal.

[0087] In this embodiment of the present invention, user input data is collected, including three types of data: text input data obtained through text boxes, voice recognition transcription, and other methods; image input data obtained through user-uploaded images or real-time camera capture; and voice input data converted from microphone input into text using voice recognition technology. Natural language processing (NLP) technology is used to extract key information such as keywords, entities, and emotions from the text input data. Computer vision technologies (such as convolutional neural networks) are used to extract key features such as objects, scenes, colors, and shapes from images. Speech recognition and audio processing technologies are used to extract keywords, emotional tones, and emotional states from speech. For text, image, and voice data, a unified semantic representation technology (such as a multimodal fusion model based on deep learning) is used to map information from different modalities into a common semantic space. This semantic space will unify the semantics of various data types, allowing information from different modalities to work together and complement each other. If any input data contains missing values, the system will issue a warning and prompt the user to supplement them. For input data with inconsistent formats, the system will automatically convert the format, such as converting non-standard date formats to a unified format or cleaning special characters in text. For duplicate data in user input, the system will automatically detect and remove redundant duplicates to ensure the uniqueness and accuracy of the data.

[0088] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes:

[0089] Step S21: Perform historical search information retrieval on the standard user input data. When historical search records are found, retrieve the user's historical search data, and construct a user profile based on the user's historical search data to generate an accurate user search profile.

[0090] Step S22: When no existing historical records are retrieved, user intent recognition is performed on the standard user input data to obtain user intent recognition data; the user intent recognition data is used to perform retrieval reasoning on the standard user input data to generate a user fuzzy retrieval profile;

[0091] Step S23: constructing a context graph for the standard user input data to obtain a user input context graph; dynamically updating the graph context complexity weight of the user input context graph based on the user precise retrieval profile and the user fuzzy retrieval profile to generate graph context dynamic update data;

[0092] Step S24: Calculate the correlation edge nodes of the graph context dynamic update data, and perform cross-query context inheritance on the user input context graph through the correlation edge nodes to generate similarity retrieval inheritance data.

[0093] In an embodiment of the present invention, standard user input data is obtained, which includes search keywords, user interests, browsing history, and other personal preference data. Historical retrieval queries are performed using the standard user input data. Historical data is stored in a database (such as the user's previous search records, behavior data, known preferences, etc.). When there are historical retrieval records in the query results, the relevant historical data is retrieved and analyzed, and these data will be used to build a user profile. Based on the historical retrieval data, machine learning or statistical models are used to generate a precise user retrieval profile, which specifically reflects the user's long-term interests, search behavior, and historical query information related to the user. If no relevant information is retrieved in the historical records, user intent is identified based on the standard user input data. Intent identification can use natural language processing (NLP) methods, such as intent classification and sentiment analysis, to identify the user's basic needs (for example, the user is looking for a product, service, information or answer). By analyzing the user's intent, an inference mechanism is used to infer the standard user input data, and a user fuzzy retrieval profile is generated based on the inference results. This can be a preliminary profile based on the user's current input and predicted needs. Based on standard user input data and its corresponding contextual information (such as time, location, and user environment), a user input context graph is constructed. This step involves converting the input data and contextual information (e.g., the time, location, and device of the user's query) into a graph structure, which can be implemented using a graph database or graph neural network. The graph context complexity weight is dynamically updated using the user's precise and fuzzy retrieval profiles. This step aims to dynamically adjust the graph structure and the weights of each node and edge based on the user's input and profile to reflect the user's current focus, interests, and query purpose. Based on the dynamically updated graph context data, the correlation edge nodes between each node and edge are calculated. This step can be used to calculate the relationship and importance between nodes using graph algorithms (such as PageRank and HITS). Using the calculated correlation edge nodes, cross-query context inheritance is performed, which inherits related query history or similar contextual information into new queries. This generates similarity retrieval inheritance data, allowing users to leverage previous query information for more precise searches in subsequent searches.

[0094] Preferably, step S22 includes the following steps:

[0095] Step S221: When no existing historical records are retrieved, keyword part-of-speech extraction is performed on the standard user input data to obtain keyword part-of-speech extraction data; sentiment classification is performed on the keyword part-of-speech extraction data to generate keyword sentiment classification data;

[0096] Step S222: performing a tendency analysis on the standard user input data using the keyword sentiment classification data to generate user intention recognition data; performing intent classification on the user intention recognition data to obtain user intention classification data;

[0097] Step S223: performing behavior matching based on the user intention classification data to obtain user behavior data; performing user preference association inference based on the user behavior data to obtain user preference inference data;

[0098] Step S224: Cluster users based on user preference inference data to obtain user fuzzy retrieval portraits.

[0099] In this embodiment of the present invention, part-of-speech tagging is performed on standard user input data, and natural language processing (NLP) techniques are used to analyze the part-of-speech of each word. For example, dependency parsing tools are used to extract nouns, verbs, adjectives, and other elements from the input data. Next, representative and high-frequency keywords are screened to form "keyword part-of-speech extraction data." For example, verbs or nouns are highly important to user intent. Sentiment analysis is performed on the extracted keywords. Using a pre-trained sentiment classification model (such as LSTM or BERT), the sentiment tendency (positive, negative, or neutral) of the user input data is determined. The analysis results are then converted into "keyword sentiment classification data," providing a preliminary understanding of user emotions and supporting subsequent user intent identification and behavior matching. Using this sentiment classification data, a tendency analysis is performed on the user input data to identify the user's emotional tendencies and underlying intentions. For example, positive sentiment indicates interest, while negative sentiment represents complaint or dissatisfaction. By combining sentiment and part-of-speech, "user intent identification data" is generated, describing the emotion and goal of the user input. User intent is classified based on the intent characteristics in the user input. For example, based on sentiment analysis and key term recognition, user intent can be categorized into "purchase intent," "query intent," "complaint intent," and so on. This generates "user intent classification data," providing a basis for subsequent behavior matching and personalized service recommendations. Based on the user intent classification data, corresponding historical behavior data can be matched. Historical user behavior data (such as click history, search history, and purchase history) can be leveraged to identify behavioral patterns related to the current intent. "User behavior data" is generated through algorithms (such as collaborative filtering and content recommendation algorithms) to understand the user's current needs. Based on user behavior data, machine learning or data mining techniques (such as decision trees and cluster analysis) are used to infer the user's underlying preferences, generating "user preference inference data" that provides precise basis for personalized recommendation systems, advertising delivery, and customized user services. Based on "user preference inference data," users are clustered. Unsupervised learning algorithms (such as K-means and DBSCAN) are used to classify users into different categories. Fuzzy matching is performed across different user categories to construct a "user fuzzy retrieval profile" that not only reflects the user's explicit preferences but also their underlying interests and needs.

[0100] Preferably, step S24 includes the following steps:

[0101] Step S241: Performing initial node screening on the user input context graph based on the graph context dynamic update data to obtain initial screening nodes; performing boundary point relevance scoring on the initial screening nodes using a graph algorithm to obtain a relevance score for each initial screening node;

[0102] Step S242: Filter out the edge nodes with the highest correlation according to the correlation score of each initial screening node, and mark them as correlation edge nodes; filter the edge-connected nodes of the initial screening nodes through the correlation edge nodes to obtain adjacent associated screening nodes;

[0103] Step S243: performing node directionality analysis on the adjacent associated screening nodes to generate adjacent associated screening node directionality data; performing context node marking on the adjacent associated screening nodes based on the adjacent associated screening node directionality data to obtain previous associated nodes and following associated nodes;

[0104] Step S244: Calculate the node information entropy value of the correlation edge node through the previous related node and the following related node to obtain the standard node information entropy value; perform cross-query context inheritance on the user input context graph through the standard node information entropy value to generate similarity retrieval inheritance data.

[0105] In an embodiment of the present invention, the user input context graph is initially screened for nodes by dynamically updating data based on the graph context. By analyzing the user's contextual data (e.g., the user's search terms, input text, behavioral data, etc.), relevant initial nodes in the graph are extracted. A graph algorithm (e.g., a graph neural network, Dijkstra algorithm, etc.) is used to score the boundary point relevance of the initial screening nodes, measuring the relevance of each initial node to the input context. For example, the relevance can be evaluated by calculating the similarity between nodes (e.g., cosine similarity, Euclidean distance, etc.), resulting in a relevance score for each initial screening node. Based on the relevance score, the nodes most relevant to the user input context are selected to obtain a set of "initial screening nodes" as the basis for further analysis. Based on the relevance score of each initial screening node, the edge nodes with the highest relevance (i.e., the nodes with the strongest association with the initial screening node) are screened. These edge nodes represent other nodes closely related to the initial node and are the key to further expanding the graph. These screened edge nodes are marked as "relevance edge nodes" to provide a basis for subsequent association analysis. Based on the screened relevance edge nodes, the initial screening nodes are screened for edge-connected nodes. Nodes connected by correlation edge nodes form "adjacent correlation filtering nodes." These nodes have direct or indirect relationships with the initial filtering nodes in the graph. Node directionality analysis is performed on these adjacent correlation filtering nodes, analyzing their structural position and relationship direction (e.g., parent-child, previous-next, causal, etc.) in the graph. For example, directed edges in the graph are used to analyze the relative directionality of nodes, generating "adjacent correlation filtering node directionality data." This records directional information for each adjacent node, such as whether it is a "previous" node (i.e., a causal, historical, or temporal upward relationship with the current node) or a "subsequent" node (i.e., an influencing or subsequent relationship with the current node). Based on this directional data, adjacent correlation filtering nodes are contextually labeled. This step can be performed by setting specific rules (e.g., temporal order, semantic association, etc.) to label nodes as "previous" or "subsequent" nodes. This labeling allows nodes in the graph to be differentiated and categorized based on contextual relationships. By analyzing both previous and subsequent nodes, the node information entropy of the correlation edge nodes is calculated. Information entropy can be used to measure the amount of information and uncertainty associated with a node in a graph. Generally, higher node entropy values ​​indicate richer information and greater uncertainty. Standard node entropy values ​​are calculated to assess the importance of graph nodes and their role within the graph. Cross-query context inheritance is performed based on standard node entropy values. This means that the entropy values ​​of existing nodes are used to infer and match new user query contexts, ensuring that relevant contextual information is inherited from the graph.Through the inheritance process, "similarity retrieval inheritance data" is generated, that is, based on information entropy and node relevance, the nodes and information related to the user query in the graph are expanded to provide more accurate context support for subsequent retrieval.

[0106] Preferably, performing cross-query context inheritance on the user input context graph by using node information entropy values ​​includes:

[0107] Perform context topology analysis on the user input context graph to obtain context structure topology data; perform semantic inheritance relationship extraction on the context structure topology data to generate semantic inheritance relationship data;

[0108] Perform query pattern matching on semantic inheritance relationship data through node information entropy value to generate query pattern matching data; perform propagation path analysis on user input context graph based on query pattern matching data to generate context query propagation path data;

[0109] The standard node information entropy value is used to verify the cross-node inheritance rules of the context query propagation path data. When the information entropy value of the node in the context query propagation path is less than or equal to the standard node information entropy value, similarity retrieval inheritance data is generated; otherwise, the corresponding context query propagation path is eliminated.

[0110] In an embodiment of the present invention, a topological structure analysis is performed on the context graph input by the user. Through a graph algorithm, the relationships and connection patterns between the nodes in the graph, such as parent-child relationships, sibling relationships, dependency relationships, etc., are identified to generate "context structure topological data" to record each node in the graph and its connection method. This topological data may include the hierarchical relationship of the nodes, adjacent node information and the weight of the edges between the nodes. Based on the context structure topological data, the semantic inheritance relationship is extracted. The semantic inheritance relationship refers to the inheritance and extension of the nodes in the context graph at the semantic level. For example, a node represents a concept, and its related child nodes inherit certain attributes or semantics of the concept, generating "semantic inheritance relationship data", including the semantic connection between the node and its directly or indirectly related nodes. These data help to understand the actual meaning of the nodes in the context graph and the relationship between them. The node information entropy value is used to perform query pattern matching on the "semantic inheritance relationship data". Based on the user's query or contextual information, the system matches relevant node patterns in the graph to ensure that the nodes in the graph closely align with the user's query intent and requirements. This generates "query pattern matching data," including all nodes that meet the query criteria and their relationships. This data is used in subsequent context propagation path analysis to ensure query accuracy and relevance. Based on this generated query pattern matching data, the propagation path of the user's input context graph is analyzed. Propagation path analysis begins with the initial node entered by the user and analyzes how information flows and propagates within the graph based on connections and inheritance relationships within the graph. This generates "context query propagation path data," which records the propagation order and related information of each node and path during the query process. This helps identify which paths best meet the query requirements. Cross-node inheritance rules are validated against this "context query propagation path data" using standard node entropy values. Each node in the propagation path is compared based on its entropy value against the standard node entropy value. If a node's entropy value is less than or equal to the standard node entropy value, it is considered highly reliable in the context propagation path and can be retained. If the information entropy value of a node is greater than the standard node information entropy value, it is considered that the node has low information content or high noise in the propagation path, resulting in inaccurate retrieval, and should be eliminated. If the node in the propagation path meets the inheritance rules (that is, the node information entropy value is less than or equal to the standard node information entropy value), the node will continue to be retained and "similarity retrieval inheritance data" will be generated. This data includes the nodes and paths that are most relevant to the user's query, and can provide high-quality, personalized retrieval results. If the node information entropy value of a certain propagation path does not meet the rules, the corresponding context query propagation path will be eliminated. This process ensures the accuracy and effectiveness of the node information in the query path.

[0111] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes:

[0112] Step S31: using an external domain knowledge base to obtain external information data;

[0113] Step S32: performing a cross-domain similarity search on the user input data according to the similarity search inheritance data to obtain cross-domain similarity search data; performing a search field analysis on the cross-domain similarity search data to generate search field analysis data;

[0114] Step S33: performing dynamic knowledge retrieval on the cross-domain similar retrieval data based on the retrieval domain analysis data to generate dynamic knowledge retrieval data; performing multi-knowledge source fusion on the dynamic knowledge retrieval data and the cross-domain similar retrieval data to generate fused retrieval data;

[0115] Step S34: Perform user input context sentiment analysis on the standard user input data through a preset sentiment analysis model to generate user input context sentiment data; use the user input context sentiment data to perform retrieval display adjustment on the fused retrieval data to generate the final hybrid retrieval display data.

[0116] In an embodiment of the present invention, external information related to the user input data is obtained by utilizing external domain knowledge bases (such as academic databases, industry-specific data sources, network resources, open APIs, etc.). These knowledge bases can be structured (such as databases) or unstructured (such as text, articles, reports, etc.). Relevant information is extracted from the external knowledge base using a query mechanism or API call, and matched and filtered in combination with the current user input data to obtain relevant "external information data". External information data can include expert advice, industry reports, historical data, research literature, etc., providing basic information for subsequent similarity retrieval and knowledge fusion. Based on the "similarity retrieval inheritance data", the user input data is subjected to cross-domain similarity retrieval. This step uses known cross-domain data mapping, semantic matching and other technologies to compare the user input data with data in multiple fields to find similar data items. For example, the user's query involves multiple fields (such as medicine and engineering). Through cross-domain technologies (such as transfer learning, semantic mapping, etc.), information similar to the user's query is found from knowledge bases in different fields. The obtained "cross-domain similarity retrieval data" is subjected to retrieval domain analysis. Specifically, the search results are analyzed to determine the domain to which they belong, identifying the domain label and relevant attributes for each search data point. This generates "search domain analysis data," which includes information such as the domain, category, and relevance score for each search result. This step helps understand the diversity of search results and which domains' information is most relevant. Based on this "search domain analysis data," dynamic knowledge retrieval is performed on "cross-domain similar search data." By dynamically adjusting query methods (e.g., through changes in query syntax, adjustments to real-time data feedback, and contextual reasoning), more domain knowledge relevant to the user's query is acquired, generating "dynamic knowledge retrieval data." This data includes knowledge points that are further optimized based on user needs and highly relevant to the user input. Multi-knowledge source fusion is then performed on the "dynamic knowledge retrieval data" and "cross-domain similar search data." Fusion can be performed through weighted fusion, priority ranking, and model fusion to integrate knowledge from multiple data sources, eliminate information conflicts, and enhance data comprehensiveness and accuracy. This generates "fused search data," which integrates search results and knowledge from multiple domains and assigns weighted scores based on the relevance of each domain. Using a pre-set sentiment analysis model, sentiment analysis of the user input context is performed on "standard user input data." Sentiment analysis models can be based on machine learning (such as deep learning models and support vector machines) or rule-based sentiment classification models. They analyze user emotions, intentions, preferences, and generate "user input contextual sentiment data," including the user's emotional tendency (such as positive, negative, neutral), emotional intensity, and emotional type (such as joy, anger). Based on this "user input contextual sentiment data," the display of the "fused search data" is adjusted. The display of search results is adjusted based on the user's emotional tendency.For example, when a user expresses positive emotions, priority is given to displaying search results that are highly relevant, have good feedback, and are more positive; when a user expresses negative emotions, more soothing and suggestive information is provided to generate the "final mixed search display data," which includes search results that have been adjusted for emotions, aiming to improve user satisfaction and optimize the user experience.

[0117] Preferably, performing user input context sentiment analysis on standard user input data using a preset sentiment analysis model includes:

[0118] Performing sentiment dictionary matching on the standard user input data to generate sentiment dictionary matching data; performing sentiment vocabulary extraction on the standard user input data based on the sentiment dictionary matching data to obtain sentiment vocabulary data;

[0119] Perform sentiment polarity analysis based on sentiment vocabulary data and determine the sentiment tendency in standard user input data to obtain sentiment polarity data; perform contextual sentiment enhancement on the sentiment polarity data using a preset sentiment analysis model to generate contextual sentiment enhancement data;

[0120] Contextual emotion enhancement data is used to evaluate the emotion intensity of standard user input data and generate user input context emotion data.

[0121] In an embodiment of the present invention, a suitable sentiment dictionary is selected, such as the commonly used SentiWordNet, a sentiment dictionary (such as the Chinese Baidu sentiment dictionary, or other sentiment dictionaries for specific fields). The sentiment dictionary contains a large number of words with sentiment attributes, and each word is accompanied by a sentiment label (such as "positive", "negative", "neutral") and its intensity value. The vocabulary in the standard user input data is matched with the entries in the sentiment dictionary. Through natural language processing technology (such as word segmentation, part-of-speech tagging, etc.), the keywords in the input data are extracted, and then compared with the sentiment dictionary to find the sentiment words therein. After matching, sentiment dictionary matching data is generated, which includes all matched sentiment words in the input data and their corresponding sentiment polarity labels (such as "positive", "negative") and intensity (such as strong, medium, weak). Based on the matched sentiment dictionary matching data, emotion-related words are extracted from the standard user input data. Sentiment vocabulary includes words that express emotions, feelings, attitudes, evaluations, etc., such as "happy", "sad", "satisfied", etc., and sentiment vocabulary data is generated, including all sentiment words in the input data and their contextual positions in the sentence. The extracted sentiment words are filtered to remove irrelevant words (such as neutral words and non-emotional function words). Part-of-speech tagging and semantic analysis are used to ensure that the extracted sentiment words truly reflect the emotional state of the user input. Sentiment polarity analysis is performed based on the extracted sentiment word data. Sentiment polarity analysis is the process of determining sentiment tendency, that is, determining whether a sentiment word belongs to "positive" or "negative" sentiment. Sentiment analysis algorithms (such as Naive Bayes, Support Vector Machines, and deep learning models) can be used to classify sentiment words and generate sentiment polarity data. Sentiment polarity data includes the polarity (positive, negative, or neutral) and intensity of each sentiment word. Based on the polarity information of sentiment words, the sentiment tendency of the entire standard user input data is determined. Typically, the sentiment tendency of the input data is determined by taking a weighted average of the sentiment words. If the majority of sentiment words lean toward "positive," the overall sentiment is positive, and vice versa. The resulting sentiment polarity data includes a categorical label for the sentiment tendency (positive, negative, mixed, etc.) and the intensity of the sentiment (such as strong, moderate, or mild). In sentiment analysis, the sentiment value of a single sentiment word varies depending on the context, so sentiment enhancement requires contextual consideration. Contextual sentiment enhancement uses contextual information to adjust the sentiment intensity and polarity of sentiment words. For example, if a sentiment word has a stronger sentiment value in certain contexts, its weight can be increased based on the context. Sentiment polarity data is enhanced using a preset sentiment analysis model (such as context-aware deep learning models like LSTM and BERT). The weight of sentiment words is increased or adjusted through contextual relationship analysis. Contextual sentiment enhancement generates enhanced contextual sentiment data, which includes the sentiment intensity and polarity adjusted after contextual analysis.For example, in positive sentiment sentences, the positive intensity of sentiment words is enhanced; in negative sentiment sentences, the intensity of negative sentiment is enhanced. Based on the contextual sentiment enhancement data, the standard user input data is evaluated for sentiment intensity. This step is the final quantification and evaluation of the sentiment tendency and sentiment intensity in the user input data. Sentiment intensity evaluation can use weighted averaging methods, sentiment analysis models (such as deep neural networks), etc. to perform weighted processing on each sentiment word in the input data, and finally obtain the comprehensive sentiment intensity of the entire input data. Based on the results of the sentiment intensity evaluation, the final user input context sentiment data is generated. This data contains detailed evaluation results of the sentiment tendency (such as positive, negative, mixed, etc.) and sentiment intensity (such as strong, medium, weak) of the user input. This sentiment data can reflect the user's emotional state in a specific context, and provide important emotional information basis for subsequent recommendations, feedback, etc.

[0122] Preferably, step S4 includes the following steps:

[0123] Step S41: collecting user consultation feedback on the final hybrid search display data to obtain user consultation feedback data, wherein the user consultation feedback collection includes click feedback collection and dwell time feedback collection;

[0124] Step S42: performing hybrid search optimization on the final hybrid search display data based on the user consultation feedback data to perform information technology search consultation optimization operations.

[0125] In this embodiment of the present invention, two main types of feedback data are collected during user interaction with the system: The first is to record user click behavior on different items in the search results, including the content clicked, the order of clicks, and the frequency of clicks. Click feedback reflects the user's attention and interest in certain information and is an important basis for analyzing user needs and optimizing search results. The second is to record the time a user spends on each search result or page. A longer dwell time indicates a user's interest in the content or a need for more time to understand it. This data can help analyze which content is most attractive or useful to users. Click feedback is captured using the front-end interface and user interaction tools (such as buttons, links, page jumps, and pop-ups). Page analysis tools (such as Google Analytics or custom page tracking scripts) are used to capture user dwell time. These tools record the dwell time on each page or content, providing data support for subsequent analysis. The collected click feedback and dwell time feedback data requires preprocessing and cleaning to remove invalid and duplicate records and ensure data accuracy and validity. The data is formatted to ensure that it conforms to the data structure required for subsequent analysis and facilitates subsequent processing. The "final hybrid search display data" is optimized based on user click feedback and dwell time feedback. This step aims to dynamically adjust search results based on real-world user behavior to improve relevance and user experience. By analyzing the frequency and order of user clicks, we can identify which results are of greatest interest to users. These frequently clicked results are prioritized or weighted higher in future searches. Items with high click counts are considered more appealing to users and thus given more exposure in subsequent searches. Based on the time users spend on a page, we analyze which content retains users for extended periods. This content can be displayed more frequently or presented in a more refined format (such as by adding summary information or recommendation tags) to enhance its visibility and appeal. Based on the integration of multiple knowledge sources, the search algorithm is adjusted in conjunction with user behavior data, employing a weighted model based on user feedback. The weights and ranking rules of individual search result items are dynamically adjusted based on user click-through rate, dwell time, and other data. By introducing reinforcement learning or online learning models, we continuously receive and analyze user feedback data to optimize the search display strategy in real time. This approach allows the system to gradually adapt to changing user interests, improving the accuracy and personalization of search results. Based on the optimized hybrid search results, information technology search and consultation optimization tasks are performed. Specifically: in the information retrieval system, database indexes and query algorithms are re-adjusted to prioritize the optimized content during query presentation; in the user consultation interface, the display strategy is adjusted, and search terms are re-ordered based on the optimization results, so that users first see results that match their preferences; and based on user consultation feedback, node weights in the knowledge graph are adjusted to reflect users' actual needs and concerns.

[0126] In this specification, an information technology search and consultation system is provided for executing the above-mentioned information technology search and consultation method. The information technology search and consultation system includes:

[0127] The data processing module is used to obtain user input data; extract the modal key information of the user input data to obtain multimodal key semantic information data; perform data preprocessing on the multimodal key semantic information data to generate standard user input data;

[0128] The retrieval inheritance module is used to retrieve historical retrieval information of standard user input data. When historical retrieval records are retrieved, a precise portrait of the standard user input data is constructed to generate a precise retrieval portrait of the user. When no historical records are retrieved, retrieval reasoning is performed on the standard user input data to generate a fuzzy retrieval portrait of the user. A user input context map is constructed, and cross-query context inheritance is performed on the user input context map based on the precise retrieval portrait and the fuzzy retrieval portrait to generate similarity retrieval inheritance data.

[0129] The fusion retrieval module is used to obtain external information data; perform dynamic knowledge retrieval on user input data based on similarity retrieval inheritance data to generate dynamic knowledge retrieval data; perform multi-knowledge source fusion on dynamic knowledge retrieval data and cross-domain similarity retrieval data to generate fusion retrieval data; perform retrieval display adjustment on the fusion retrieval data through a preset sentiment analysis model to generate the final hybrid retrieval display data;

[0130] The retrieval feedback module is used to perform hybrid retrieval feedback optimization on the final hybrid retrieval display data to perform information technology retrieval consulting optimization operations.

[0131] The beneficial effect of the present invention is that by obtaining user input and extracting multimodal key information, different forms of user input (such as text, images, voice, etc.) are converted into standardized data formats to ensure the consistency and processability of the input data. This preprocessing process helps the subsequent retrieval system to understand user needs more accurately and provide a reliable data foundation for accurate retrieval and reasoning. By generating accurate or fuzzy retrieval portraits of users through historical retrieval records and reasoning, personalized demand analysis is achieved. When there are no historical records, fuzzy portraits are generated through reasoning to ensure that the system can handle new users or small amounts of data. At the same time, through cross-query context inheritance technology, the relevance in continuous queries is improved, the context consistency during multiple interactions is guaranteed, and the accuracy of retrieval is further enhanced by similarity retrieval inheritance data. Through dynamic knowledge retrieval and cross-domain information fusion, relevant information can be obtained from multiple knowledge sources, improving the breadth and depth of retrieval. Multi-knowledge source fusion avoids the problem of information islands and increases the system's coverage of multi-domain data. At the same time, the introduction of the sentiment analysis model enables the retrieval results to be adjusted according to the user's emotional tendencies, thereby improving the personalization of the retrieval results and user experience. Through retrieval feedback optimization, the module can dynamically adjust the retrieval strategy and results according to user feedback, thereby improving the intelligence and adaptability of the retrieval system. The continuous optimization process ensures that the system is constantly improved, improves the relevance, accuracy and real-time performance of query results, and enhances the learning ability of the system so that future retrieval can better meet user needs. Therefore, the present invention solves the problems of low retrieval accuracy, incomplete information coverage and poor user experience in the prior art by combining multimodal information extraction, precise and fuzzy portrait construction, cross-context inheritance, dynamic knowledge fusion and sentiment analysis optimization.

[0132] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0133] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. An information technology search and consultation method, characterized in that: The following steps are involved: Step S1: obtaining user input data; extracting modal key information of the user input data to obtain multimodal key semantic information data; performing data preprocessing on the multimodal key semantic information data to generate standard user input data; Step S2: Perform historical search information retrieval on the standard user input data. If a historical search record is found, a precise profile is constructed for the standard user input data to generate a precise user search profile. If no historical record is found, a search reasoning is performed on the standard user input data to generate a fuzzy user search profile. Construct a user input context map, and perform cross-query context inheritance on the user input context map based on the user's precise retrieval profile and the user's fuzzy retrieval profile to generate similarity retrieval inheritance data; wherein step S2 includes the following steps: Step S21: Perform historical search information retrieval on the standard user input data. When historical search records are found, retrieve the user's historical search data, and construct a user profile based on the user's historical search data to generate an accurate user search profile. Step S22: When no existing historical records are retrieved, user intent recognition is performed on the standard user input data to obtain user intent recognition data; the user intent recognition data is used to perform retrieval reasoning on the standard user input data to generate a user fuzzy retrieval profile; Step S23: constructing a context graph for the standard user input data to obtain a user input context graph; dynamically updating the graph context complexity weight of the user input context graph based on the user precise retrieval profile and the user fuzzy retrieval profile to generate graph context dynamic update data; Step S24: Calculate the correlation edge nodes of the graph context dynamic update data, and perform cross-query context inheritance on the user input context graph through the correlation edge nodes to generate similarity retrieval inheritance data; wherein, step S24 includes the following steps: Step S241: Performing initial node screening on the user input context graph based on the graph context dynamic update data to obtain initial screening nodes; performing boundary point relevance scoring on the initial screening nodes using a graph algorithm to obtain a relevance score for each initial screening node; Step S242: Filter out the edge nodes with the highest correlation according to the correlation score of each initial screening node, and mark them as correlation edge nodes; filter the edge-connected nodes of the initial screening nodes through the correlation edge nodes to obtain adjacent associated screening nodes; Step S243: performing node directionality analysis on the adjacent associated screening nodes to generate adjacent associated screening node directionality data; performing context node marking on the adjacent associated screening nodes based on the adjacent associated screening node directionality data to obtain previous associated nodes and following associated nodes; Step S244: Calculate the node information entropy value of the correlation edge node through the previous associated node and the following associated node to obtain the standard node information entropy value; perform cross-query context inheritance on the user input context map through the standard node information entropy value to generate similarity retrieval inheritance data; wherein, performing cross-query context inheritance on the user input context map through the node information entropy value includes: Perform context topology analysis on the user input context graph to obtain context structure topology data; perform semantic inheritance relationship extraction on the context structure topology data to generate semantic inheritance relationship data; Perform query pattern matching on semantic inheritance relationship data through node information entropy value to generate query pattern matching data; perform propagation path analysis on user input context graph based on query pattern matching data to generate context query propagation path data; The cross-node inheritance rule of context query propagation path data is verified using the standard node information entropy value. When the information entropy value of the node in the context query propagation path is less than or equal to the standard node information entropy value, similarity retrieval inheritance data is generated; otherwise, the corresponding context query propagation path is eliminated. Step S3: Acquire external information data; perform dynamic knowledge retrieval on user input data based on similarity retrieval inheritance data to generate dynamic knowledge retrieval data; perform multi-knowledge source fusion on the dynamic knowledge retrieval data and cross-domain similarity retrieval data to generate fused retrieval data; perform retrieval display adjustment on the fused retrieval data using a preset sentiment analysis model to generate final hybrid retrieval display data; Step S4: Perform hybrid search feedback optimization on the final hybrid search display data to perform information technology search consulting optimization operations.

2. The information technology search and consultation method according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: obtaining user input data, wherein the user input data includes text input data, image input data and voice input data; Step S12: extracting modal key information from the text input data, the image input data, and the voice input data to obtain multimodal key information extraction data; Step S13: performing unified semantic representation on the multimodal key information extraction data to generate multimodal key semantic information data; Step S14: preprocess the multimodal key semantic information data to generate standard user input data, wherein the data preprocessing includes data missing value prompts, data abnormal format conversion, and data duplicate value removal.

3. The information technology search and consultation method according to claim 1, characterized in that: Step S22 includes the following steps: Step S221: When no existing historical records are retrieved, keyword part-of-speech extraction is performed on the standard user input data to obtain keyword part-of-speech extraction data; sentiment classification is performed on the keyword part-of-speech extraction data to generate keyword sentiment classification data; Step S222: performing a tendency analysis on the standard user input data using the keyword sentiment classification data to generate user intention recognition data; performing intent classification on the user intention recognition data to obtain user intention classification data; Step S223: performing behavior matching based on the user intention classification data to obtain user behavior data; performing user preference association inference based on the user behavior data to obtain user preference inference data; Step S224: Cluster users based on user preference inference data to obtain user fuzzy retrieval portraits.

4. The information technology search and consultation method according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: using an external domain knowledge base to obtain external information data; Step S32: performing a cross-domain similarity search on the user input data according to the similarity search inheritance data to obtain cross-domain similarity search data; performing a search field analysis on the cross-domain similarity search data to generate search field analysis data; Step S33: performing dynamic knowledge retrieval on the cross-domain similar retrieval data based on the retrieval domain analysis data to generate dynamic knowledge retrieval data; performing multi-knowledge source fusion on the dynamic knowledge retrieval data and the cross-domain similar retrieval data to generate fused retrieval data; Step S34: Perform user input context sentiment analysis on the standard user input data through a preset sentiment analysis model to generate user input context sentiment data; use the user input context sentiment data to perform retrieval display adjustment on the fused retrieval data to generate the final hybrid retrieval display data.

5. The information technology search and consultation method according to claim 4, characterized in that: The user input context sentiment analysis of standard user input data using a preset sentiment analysis model includes: Performing sentiment dictionary matching on the standard user input data to generate sentiment dictionary matching data; performing sentiment vocabulary extraction on the standard user input data based on the sentiment dictionary matching data to obtain sentiment vocabulary data; Perform sentiment polarity analysis based on sentiment vocabulary data and determine the sentiment tendency in standard user input data to obtain sentiment polarity data; perform contextual sentiment enhancement on the sentiment polarity data using a preset sentiment analysis model to generate contextual sentiment enhancement data; Contextual emotion enhancement data is used to evaluate the emotion intensity of standard user input data and generate user input context emotion data.

6. The information technology search and consultation method according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: collecting user consultation feedback on the final hybrid search display data to obtain user consultation feedback data, wherein the user consultation feedback collection includes click feedback collection and dwell time feedback collection; Step S42: performing hybrid search optimization on the final hybrid search display data based on the user consultation feedback data to perform information technology search consultation optimization operations.

7. An information technology retrieval and consultation system, characterized in that: For executing the information technology search and consultation method according to claim 1, the information technology search and consultation system comprises: The data processing module is used to obtain user input data; extract the modal key information of the user input data to obtain multimodal key semantic information data; perform data preprocessing on the multimodal key semantic information data to generate standard user input data; The retrieval inheritance module is used to retrieve historical retrieval information of standard user input data. When historical retrieval records are retrieved, a precise portrait of the standard user input data is constructed to generate a precise retrieval portrait of the user. When no historical records are retrieved, retrieval reasoning is performed on the standard user input data to generate a fuzzy retrieval portrait of the user. A user input context map is constructed, and cross-query context inheritance is performed on the user input context map based on the precise retrieval portrait and the fuzzy retrieval portrait to generate similarity retrieval inheritance data. The fusion retrieval module is used to obtain external information data; perform dynamic knowledge retrieval on user input data based on similarity retrieval inheritance data to generate dynamic knowledge retrieval data; perform multi-knowledge source fusion on dynamic knowledge retrieval data and cross-domain similarity retrieval data to generate fusion retrieval data; perform retrieval display adjustment on the fusion retrieval data through a preset sentiment analysis model to generate the final hybrid retrieval display data; The retrieval feedback module is used to perform hybrid retrieval feedback optimization on the final hybrid retrieval display data to perform information technology retrieval consulting optimization operations.

Images