Intelligent query system and method based on semantic analysis

Abstract

The invention relates to the technical field of information retrieval, in particular to an intelligent query system and method based on semantic analysis, and the system comprises an entity relation mapping module which collects user query data, carries out part-of-speech tagging and syntactic analysis operation on user query, constructs connection mapping according to syntactic relations among key elements, and sends the connection mapping to an entity query module; and calculating the correlation degree between each pair of keywords in the mapping, quantifying the attribute interaction strength, and generating a keyword relation mapping table. According to the method and the device, the precision and the efficiency of information retrieval are improved by deeply analyzing the query data of the user and combining the technical innovation of natural language processing. When part-of-speech tagging and syntactic analysis are carried out on a user query statement, key elements are accurately identified, connection mapping is constructed, and accurate capture of core information is ensured. And the information retrieval precision is enhanced by quantifying the association degree between the keywords and the attribute interaction strength. And a coherence score is calculated, so that the system can screen out the most relevant information.

Description

Technical Field

[0001] The present invention relates to the field of information retrieval technology, and in particular to an intelligent query system and method based on semantic analysis. Background Art

[0002] Information retrieval is a technical field that involves automatically retrieving, organizing, and searching information stored on electronic devices. It is also a problem of finding relevant information from a large number of unstructured data sources (such as web pages, books, journal articles, and multimedia files). It aims to improve the efficiency of information search and the relevance of results, and is widely used in search engines, digital libraries, and enterprise management systems.

[0003] Among them, the intelligent query system processes and responds to user queries in an intelligent way. The system uses natural language processing technology to understand the user's query intent and quickly and accurately extract information from large data sets. Its uses include but are not limited to providing instant search results, helping companies analyze customer data, supporting academic research, and enhancing the user experience of e-commerce websites. By intelligently analyzing query content and context, the accuracy and efficiency of information retrieval can be improved.

[0004] When processing user queries, existing technologies have difficulty in fully capturing implicit information and contextual associations in user queries. This results in low accuracy and consistency in information retrieval. When using the system, users often receive a large amount of redundant information that is not completely relevant to the query intent, which reduces the efficiency of retrieval and also prevents users from quickly finding the required information. At the same time, existing technologies tend to ignore the temporal correlation between information, resulting in insufficient information update speed and accuracy. It is difficult to achieve rapid dynamic adjustment of query parameters, which makes the system limited in response flexibility when facing complex queries and difficult to provide the best retrieval results based on real-time changes in user needs. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an intelligent query system and method based on semantic analysis.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: The intelligent query system based on semantic analysis includes:

[0007] The entity-relationship mapping module collects user query data, performs part-of-speech tagging and syntactic parsing operations on user queries, identifies key elements in query sentences, which are names of people, places, events, or specified objects, builds connection mappings based on the syntactic relationships between key elements, calculates the correlation between each pair of keywords in the mapping, quantifies the intensity of attribute interaction, and generates a keyword relationship mapping table;

[0008] The coherence analysis module receives the keyword relationship mapping table, calculates the coherence score of each connection according to the key element relationship and attribute interaction information listed in the table, sorts and filters the coherence scores, and generates a coherence score table;

[0009] The time series matching module converts the keyword connections and attribute interaction information with significant scores in the coherence scoring table and identified from the user query statement into time series data according to the time points when they appear in the text, calculates the synchronization error between the sequences, compares the matching degree of the time series, and predicts the future query trend or information demand change to generate data prediction results;

[0010] The query parameter optimization module adjusts the query parameters according to the matching degree information of the data prediction results, compares the modified query parameters with the current state of the information source, and reapplies the modified parameters to the information retrieval process to generate optimized query results.

[0011] As a further solution of the present invention, the step of obtaining the association between each pair of keywords in the calculation mapping is specifically as follows:

[0012] Collect user query data, perform word segmentation and part-of-speech tagging, identify key elements in the query, determine the syntactic dependency between elements, and generate key element recognition results;

[0013] Based on the key element recognition results, analyzing the syntactic relationship between elements, and establishing a connection mapping between key elements to generate a key element connection mapping diagram;

[0014] According to the key element connection mapping diagram, the formula is used:

[0015]

[0016] Calculate the correlation R between keywords A and B AB , considering the number of co-occurrences, grammatical weights, semantic similarity and frequency ratio, to generate keyword relevance analysis results;

[0017] Among them, C AB represents the number of times keywords A and B appear together in a query, N is the total number of queries in the corpus, and W AB is the weight coefficient based on the syntactic dependency between keywords A and B, F AB is the semantic similarity between keywords A and B, P AB is the frequency ratio of keywords A and B in the entire corpus.

[0018] As a further solution of the present invention, the step of obtaining the quantitative attribute interaction strength is specifically:

[0019] Based on the collected element interaction information, the contextual relationship of key elements is analyzed using the formula:

[0020]

[0021] Calculate the attribute interaction strength S between element i and element j ij , generating interaction strength information between elements;

[0022] Among them, A ij represents the total number of interactions between element i and element j, B ij represents the total number of elements in all interaction events in the corpus, W ij is the interaction weight coefficient, R ij is the context relevance coefficient, which indicates the degree of relevance between element i and element j in the context;

[0023] According to the interaction intensity information between the elements and the keyword association analysis results, by analyzing the syntactic structure and dependency relationship in the query statement, a relationship network diagram is constructed, all keywords and their interaction information are integrated, and a keyword relationship mapping table is generated.

[0024] As a further solution of the present invention, the steps of obtaining the coherence score sheet are specifically as follows:

[0025] Analyzing the keyword relationship in the query sentence according to the keyword relationship mapping table, evaluating the semantic relevance of the keywords in the context by calculating the coherence score between each pair of key elements, and obtaining a keyword coherence analysis result;

[0026] According to the keyword coherence analysis result, element relationships with significant coherence are screened by analyzing the scores, and a coherence score table is generated.

[0027] As a further solution of the present invention, the step of obtaining the matching degree of the comparison time series is specifically as follows:

[0028] According to the keyword connection and attribute interaction information in the coherence score table, the coherence score is brought in to mark the timestamp according to the appearance time of the keyword, and its priority in the time series is determined to generate time series data;

[0029] According to the time series data, the formula is adopted:

[0030]

[0031] Calculate the synchronization error E between two time series sync ,According to the order of time points in the time series, the differences of keywords at time points are compared, and the matching between sequences is analyzed to generate time series matching results;

[0032] Among them, X t represents the value of the time series X at the tth time point, Y t represents the value of the time series Y at the tth time point, L X and L Y Represent the length of time series X and Y, T max It represents the maximum time point value that occurs in the two time series, and n represents the total number of time points.

[0033] As a further solution of the present invention, the acquisition step of predicting future query trends or information demand changes is specifically:

[0034] According to the time series matching results, combined with the semantic relevance and keyword interaction frequency information, the formula is adopted:

[0035]

[0036] Calculate the predicted time point T pred , generate prediction time point information;

[0037] Among them, T curr is the current time point, D rel is the semantic relevance, n is the total number of time points, F int is the interaction frequency;

[0038] According to the predicted time point information, the data segments most relevant to the query trend are identified, and the weights and rankings are adjusted according to the semantic similarity of the keywords to generate data prediction results.

[0039] As a further solution of the present invention, the step of obtaining the optimized query result is specifically as follows:

[0040] According to the data prediction results, the keyword weight is evaluated, and the importance of the keyword is corrected according to the synchronization error, the query parameter is modified, and an adjusted query parameter configuration is generated;

[0041] According to the adjusted query parameter configuration, the priority and weight of the keyword are recalculated, the query task is executed, and the query result sequence is adjusted according to the keyword weight to obtain an optimized query result.

[0042] The intelligent query method based on semantic analysis is executed based on the above-mentioned intelligent query system based on semantic analysis, and comprises the following steps:

[0043] S1: Based on user query data, perform part-of-speech tagging and syntactic analysis, identify key elements, build connection mappings between key elements, calculate the correlation between keywords, and generate a keyword relationship mapping table;

[0044] S2: Based on the keyword relationship mapping table, calculate the coherence score between each keyword, sort and filter the element connections with high coherence, and generate a coherence score table;

[0045] S3: Based on the coherence score table, convert the keyword connection and interaction information into time series data, calculate the synchronization error, predict the query trend, and generate data prediction results;

[0046] S4: Based on the data prediction result, adjust the query parameters, compare the modified parameters with the current state of the information source, and generate optimized query parameters;

[0047] S5: Based on the optimized query parameters, re-execute the information retrieval task, adjust the keyword priority and the query result sequence, and generate an optimized query result.

[0048] Compared with the prior art, the advantages and positive effects of the present invention are:

[0049] In the present invention, the accuracy and efficiency of information retrieval are improved by in-depth analysis of user query data and combining the technical innovation of natural language processing. When the user query statement is tagged with parts of speech and parsed by syntax, the key elements are accurately identified and a connection mapping is constructed to ensure the accurate capture of core information. The accuracy of information retrieval is enhanced by quantifying the correlation between keywords and the intensity of attribute interaction. The calculation of the coherence score enables the system to screen out the most relevant information and ensure that the query results are highly coherent. By converting the relationship mapping of key elements into time series data, the timeliness and predictive ability of information retrieval are further optimized. This matching method based on synchronization error enables the system to respond quickly and provide expected results when dealing with complex queries, significantly improving the efficiency of information query. Through the dynamic adjustment of query parameters, continuous optimization in the information retrieval process is achieved, so that the system can automatically adjust the query strategy according to the latest demand changes, effectively improving the user experience and the intelligence of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 is a system flow chart of the present invention;

[0051] Figure 2 A flow chart of the steps for obtaining the degree of association between each pair of keywords in the mapping of the present invention;

[0052] Figure 3 A flowchart of the steps for obtaining the quantitative attribute interaction strength of the present invention;

[0053] Figure 4 is a flow chart of steps for obtaining the coherence score sheet of the present invention;

[0054] Figure 5A flow chart of the steps for obtaining the matching degree of time series comparison of the present invention;

[0055] Figure 6 A flowchart of the acquisition steps for predicting future query trends or information demand changes of the present invention;

[0056] Figure 7 A flow chart of the steps for obtaining query results optimized by the present invention. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0058] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0059] See also Figure 1 The present invention provides a technical solution: an intelligent query system based on semantic analysis, the system comprising:

[0060] The entity-relationship mapping module collects user query data, performs part-of-speech tagging and syntactic parsing operations on user queries, identifies key elements in query sentences, which are names of people, places, events, or specified objects, builds connection mappings based on the syntactic relationships between key elements, calculates the correlation between each pair of keywords in the mapping, quantifies the intensity of attribute interaction, and generates a keyword relationship mapping table;

[0061] The coherence analysis module receives the keyword relationship mapping table, calculates the coherence score of each connection according to the key element relationship and attribute interaction information listed in the table, sorts and filters the coherence scores, and generates a coherence score table;

[0062] The time series matching module converts the keyword connections and attribute interaction information that are significantly scored in the coherence score table and identified from the user query statement into time series data according to the time points at which they appear in the text. It calculates the synchronization error between sequences, compares the matching degree of the time series, and predicts future query trends or changes in information needs to generate data prediction results.

[0063] The query parameter optimization module adjusts the query parameters according to the matching information of the data prediction results, compares the modified query parameters with the current state of the information source, and reapplies the modified parameters to the information retrieval process to generate optimized query results;

[0064] The keyword relationship mapping table includes the part-of-speech categories of key elements, the dependency types between elements, and the correlation values. The coherence score table includes the coherence scores of each keyword pair, the statistical distribution of the scores, and a detailed list of high-coherence score pairs. The data prediction results include the accuracy scores of time point matching and the time deviation between sequences. The optimized query results include the optimized keyword weights, search depth adjustment values, and efficiency comparison information before and after parameter adjustment.

[0065] See also Figure 2 , the steps for obtaining the association between each pair of keywords in the calculation mapping are as follows:

[0066] Collect user query data, perform word segmentation and part-of-speech tagging, identify key elements in the query, determine the syntactic dependency between elements, and generate key element recognition results;

[0067] The system first performs part-of-speech tagging on the user's query, using a natural language processing algorithm to segment and tag the query. The system tags each word in the query as a noun, verb, or other part-of-speech category based on the context, and then parses the sentence structure through a dependency syntax analysis algorithm to identify the key elements in the query. These key elements may be names of people, places, events, or specified objects. The system determines the dependency relationships between words through syntactic structures, such as subject-predicate relationships, object relationships, etc. These key elements provide data support for connection mapping and correlation calculations in subsequent steps.

[0068] Based on the key element recognition results, the syntactic relationship between elements is analyzed, and the connection mapping between key elements is established to generate a key element connection mapping diagram;

[0069] After identifying the key elements in the query, the system analyzes the syntactic relationship between these elements through a dependency parsing algorithm, and then establishes a connection mapping between the key elements through a graph construction algorithm. The system identifies the logical connection between the elements based on the dependency relationship. For example, the relationship between a person and a place can be connected by "located in" or "occurs". The system records these relationships in a quantitative way in a mapping graph. The nodes in the mapping graph represent the key elements, and the edges represent the dependency relationships between the elements. The mapping graph provides basic data for subsequent association calculations.

[0070] According to the key element connection mapping diagram, the formula is used:

[0071]

[0072] Calculate the correlation R between keywords A and B AB , considering the number of co-occurrences, grammatical weights, semantic similarity and frequency ratio, to generate keyword relevance analysis results;

[0073] Among them, R AB Quantifies the semantic relevance between two keywords. It is an indicator used by the system to determine whether there is a strong correlation between the two keywords in terms of semantics and grammar. AB It indicates the number of times keywords A and B appear together in the query. The higher the number of co-occurrences, the stronger the semantic correlation between the two keywords. By statistically analyzing the historical user query data, the number of times keywords A and B appear together in all queries is recorded. The query log or user behavior data is collected. N is the total number of queries in the corpus, indicating the total number of all queries. It is used for normalization processing to prevent the high number of keyword co-occurrences from being misjudged as a strong correlation due to the large amount of data. The total number of all historical queries in the corpus or system is obtained. AB It is a weight coefficient based on the syntactic dependency between keywords A and B. It indicates the degree of grammatical association between the two keywords in the sentence structure. The weights are assigned according to the grammatical rules. For keywords with strong dependency, F AB It is the semantic similarity between keywords A and B, reflecting the semantic proximity of the two keywords. The semantic similarity of the two keywords is obtained through the cosine similarity analysis of the word vector model (such as Word2Vec, GloVe or BERT, etc.). The higher the value, the closer the semantics of the keywords are. AB It is the frequency ratio of keywords A and B in the entire corpus, which is obtained by counting the frequency difference of the number of times keywords A and B appear in all documents in the corpus, to prevent certain high-frequency words from affecting the accuracy of the association.

[0074] For example, keywords A and B appear together 5 times in the query, and the total number of queries in the corpus is 200. AB Based on the syntactic dependency, the semantic similarity F is 0.7. AB The value calculated by the word vector model is 0.5, and the frequency ratio difference between keywords A and B in the corpus is P. AB is 0.3, and the calculation is as follows:

[0075]

[0076] The results show that the correlation between keywords A and B is 0.0467. If the benchmark value inferred from the analysis of large-scale corpora and historical data is 0.02, any correlation exceeding 0.02 can be considered to indicate a significant correlation between the two keywords, while a correlation below this value indicates a weak correlation. This benchmark value helps the system determine the strength of correlation between keywords in actual operation and provides a reference for query optimization.

[0077] See also Figure 3 , the specific steps for obtaining the quantitative attribute interaction strength are:

[0078] Based on the collected element interaction information, the contextual relationship of key elements is analyzed using the formula:

[0079]

[0080] Calculate the attribute interaction strength S between element i and element j ij , generate interaction strength information between elements;

[0081] Among them, A ij represents the total number of interactions between element i and element j, which is obtained by counting the number of times two keywords co-occur in the same context in the corpus. Specifically, it is obtained by scanning a large amount of corpus and finding the frequency of occurrence of these keywords in the same text or query. ij Represents the total number of elements in all interaction events in the corpus, used to normalize A ij The value of W is used to prevent the total number of interactions from losing its comparative significance due to the large size of the corpus. It is obtained by counting the co-occurrence times of each key element in all queries and sentences in the corpus. ij is the interaction weight coefficient, which is used to represent the importance weight of element i and element j in different scenarios or contexts. By analyzing various scenarios, the system assigns values ​​based on the importance of keywords in different scenarios. Based on past data, the system can adjust the coefficient according to the scenario weight to better meet the query requirements. R ij is the contextual relevance coefficient, which indicates the degree of relevance between element i and element j in the context, and is obtained by analyzing different query contexts in the corpus and mining co-occurrence patterns.

[0082] For example, elements i and j appear together 10 times in the query, the total number of interaction events is 60, and the weight coefficient W ij is 0.8, and the context relevance coefficient R ij is 1.2. Substituting into the formula, the calculation is as follows:

[0083]

[0084] The results show that the interaction strength between elements i and j is 0.128. If the mean value of the interaction strength distribution in the system is 0.1 and the standard deviation is 0.05, the following intervals can be defined: high correlation: interaction strength > 0.15, medium correlation: interaction strength between 0.05 and 0.15, low correlation: interaction strength < 0.05. Therefore, 0.128 falls between 0.05 and 0.15, indicating that the interaction strength of these two elements in the corpus belongs to medium correlation, that is, the correlation is moderate.

[0085] According to the interaction intensity information between elements and the results of keyword relevance analysis, by analyzing the syntactic structure and dependency relationship in the query statement, a relationship network diagram is constructed, all keywords and their interaction information are integrated, and a keyword relationship mapping table is generated;

[0086] Integrating all the collected keyword information, the system merges each key element and its associated attribute interaction data. The system uses syntactic structure and lexical dependency to first parse the syntactic connection and interaction pattern of each key element, and then generates a global relationship network diagram through dependency parsing algorithm and graph construction algorithm. The nodes of the relationship network represent key elements, and the edges represent the dependency or interaction relationship between them. The system quantifies the interaction of each pair of keywords and generates corresponding interaction intensity values ​​through specific weights and frequency adjustment coefficients. Subsequently, the system integrates all these interaction data with keyword connections into a comprehensive relationship mapping table.

[0087] See also Figure 4 , the steps to obtain the coherence score sheet are as follows:

[0088] According to the keyword relationship mapping table, the keyword relationship in the query statement is analyzed, and the semantic relevance of the keywords in the context is evaluated by calculating the coherence score between each pair of key elements to obtain the keyword coherence analysis result;

[0089] To calculate the coherence score between each pair of key elements, the formula is used:

[0090] C ij =S ij ×R AB ×W c

[0091] Among them, C ij represents the coherence score between key elements i and j, W c It is the consistency weight coefficient, which is used to adjust the weight value of consistency calculation in different query scenarios. The coefficient is dynamically adjusted according to historical data and is usually obtained by analyzing the consistency pattern in historical queries.

[0092] If the attribute interaction strength S ij=0.128, keyword relevance R AB =0.054, coherence weight coefficient W c =1.1, then the formula is as follows:

[0093] C ij =0.128×0.054×1.1=0.00768

[0094] The coherence score between elements i and j is 0.00768. Generally, the coherence score ranges from 0 to 1. The closer the score is to 1, the stronger the coherence between the elements; the closer the score is to 0, the weaker the coherence. The results show that there is a certain degree of semantic coherence between the pair of elements.

[0095] According to the results of keyword coherence analysis, the element relationships with significant coherence are screened by analyzing the scores and a coherence score table is generated;

[0096] Based on the overall calculated coherence score C ij , sort these scores from high to low, and identify those key element connections with higher scores. During the sorting process, the system will filter out the parts with higher scores according to the set threshold. Subsequently, the filtered parts will become keyword relationships with stronger coherence. These filtered key element relationships will be recorded in the coherence score table.

[0097] See also Figure 5 , the specific steps for obtaining the matching degree of the comparison time series are:

[0098] According to the keyword connection and attribute interaction information in the coherence score table, the coherence score is brought in to timestamp the occurrence time of the keyword and determine its priority in the time series to generate time series data;

[0099] First, extract these highly coherent keywords from the user query and bring them into the calculated relevance score C ij Then, the system will match these keywords with the time points in the information source according to the order in which they appear. The system will timestamp these keywords according to their actual appearance time and combine them with the correlation C ij Determine its priority in the time series. Subsequently, the system converts these keywords and their interactive information into time series data, ensures that highly correlated keywords have a larger weight in the time series, and records their time dependencies. In this way, when the system generates time series data, it can effectively reflect the matching degree of keywords in the query statement and the information source. Finally, a structure containing time series data is generated to provide input for subsequent synchronization error calculations.

[0100] According to the time series data, the formula is used:

[0101]

[0102] Calculate the synchronization error E between two time series sync ,According to the order of time points in the time series, the differences of keywords at time points are compared, and the matching between sequences is analyzed to generate time series matching results;

[0103] Among them, X t represents the value of time series X at the tth time point. Sequence X is the time point of keywords and their interaction information extracted from user query data. It can be obtained by the order of occurrence of keywords in sentences extracted by natural language processing technology. Y t represents the value of the time series Y at the tth time point. The sequence Y is the time point of the keywords and their interactive information extracted from the information source data, which is determined by the order in which the keywords appear in the corpus or information source. X and L Y Respectively represent the length of time series X and Y, that is, the number of time points included, L X and L Y It is obtained by directly counting the total number of time points contained in the X and Y sequences and is used to correct the synchronization error of sequences of different lengths. max It represents the maximum time point value that appears in two time series. It is used for normalization processing to eliminate the influence of sequence length difference on synchronization error. It can be obtained by analyzing the timestamp information of the sequence. n represents the total number of time points, which is used to indicate how many time points are involved in the synchronization error calculation. It is determined by counting the number of all time points in the time series.

[0104] For example, the time series X is [3,5,7], the time series Y is [2,6,8], and the sequence lengths are L X =3 and L Y =3, maximum time point T max =10, substitute it into the formula and calculate as follows:

[0105]

[0106] The results show that the synchronization error between the two time series is 3.6. Generally, the closer the synchronization error is to 0, the higher the match between the two time series; the larger the value, the more obvious the difference between the time series. For example, if E sync If E is less than a lower threshold (e.g., 1.0), then the two time series can be considered to have a high degree of match, indicating that the user query and the information source are highly synchronized in terms of time and content. syncIn the middle range (e.g., 1.0 to 5.0), there is a certain degree of match between the query and the information source, but there are still large differences in time or content. sync If the error exceeds a higher threshold (e.g., above 5.0), it indicates that the match between the query and the information source is low and the synchronization between the time series is poor. In this example, the calculated error is 3.6, which belongs to the medium match range, indicating that there is some similarity between the query and the information source, but there are still some time or content differences.

[0107] See also Figure 6 ,The specific steps for predicting future query trends or changes in information needs are:

[0108] According to the time series matching results, combined with the semantic relevance and keyword interaction frequency information, the formula is used:

[0109]

[0110] Calculate the predicted time point T pred , generate prediction time point information;

[0111] Among them, T pred Represents the time point at which future query trends or information demand changes may occur, indicating the system's prediction of future trends. curr is the current time point, indicating the time point of the keyword in the current query statement in the current information source, D rel is the semantic association, which indicates the semantic relevance between keywords. Its value ranges from 0 to 1 and is obtained through natural language processing technology (such as word embedding model or semantic vector model). The larger the value, the closer the two keywords are semantically. n is the total number of time points, which indicates the total number of time points in the current time series. It is obtained by parsing the number of time points involved in the query statement and the information source. It is used to balance the accuracy of the prediction. int It is the interaction frequency, which indicates the frequency of keywords appearing in queries and information sources. It is obtained by counting the number of times the keywords appear in queries or documents. The higher the value, the greater the importance of the keyword in the query.

[0112] Setting parameter T curr =15, E sync =3.6, D rel =0.8,n=3,F int =2.

[0113] Substitute into the formula:

[0114]

[0115] The results show that the predicted future query trend time point is 15.48, which means that the system expects new query trends or changes in information needs to appear at a moment close to the current time point in the future. This time point is used to guide the system to give priority to these potential new changes in subsequent query processing, so as to meet user needs more accurately.

[0116] Based on the predicted time point information, identify the data segments most relevant to the query trend, adjust the weights and sorting based on the semantic similarity of the keywords, and generate data prediction results;

[0117] The system cleans and classifies the data in the information source, breaks it down into independent time nodes, and then compares the predicted query trends with these time nodes one by one. The system identifies the key data fragments that are most relevant to the user query at the predicted time point, and uses semantic similarity analysis to match these fragments with the keywords in the current query. Subsequently, the system adjusts the weight and sorting of keywords according to changes in time points to ensure that more important keywords are given priority during the query process. The system also combines trend changes in historical data to compare data fluctuations in future forecasts, analyzes possible abnormal trends or emerging information, and optimizes data matching logic.

[0118] See also Figure 7 , the steps for obtaining the optimized query results are as follows:

[0119] According to the data prediction results, the keyword weight is evaluated, and the importance of the keyword is corrected according to the synchronization error, the query parameters are modified, and the adjusted query parameter configuration is generated;

[0120] Based on T pred and E sync , the system adjusts the query parameters based on the matching information. First, the system adjusts the query parameters based on the predicted future time point T pred Re-evaluate the weight of keywords and use the synchronization error E sync The importance of the query keyword is corrected. If the synchronization error is large, the system will lower the priority of the keyword in future queries, otherwise it will increase its weight. The system adjusts the matching degree of the keyword at different time points to determine the query parameters that need to be modified. Subsequently, the modified query parameters are compared with the data of the current information source to ensure that the new parameter configuration is consistent with the latest status of the information source.

[0121] According to the adjusted query parameter configuration, the priority and weight of the keyword are recalculated, the query task is executed, and the query result order is adjusted according to the keyword weight to obtain an optimized query result;

[0122] Based on the adjusted query parameter configuration, the system applies these new parameters to the information query process, updates the parameter configuration, recalculates the priority and weight of each keyword, and then executes the query task according to the modified query parameters. The system will readjust the order of the query results according to the keyword weights, and filter out the information fragments that best match the latest parameters in the information source. The system will also optimize the order of the results according to the semantic relevance of the keywords to ensure that the most relevant information is ranked first, and finally generate optimized query results, which the system will use as output for users.

[0123] The intelligent query method based on semantic analysis is executed based on the above-mentioned intelligent query system based on semantic analysis, and includes the following steps:

[0124] S1: Based on user query data, perform part-of-speech tagging and syntactic analysis, identify key elements, build connection mappings between key elements, calculate the correlation between keywords, and generate a keyword relationship mapping table;

[0125] S2: Based on the keyword relationship mapping table, calculate the coherence score between each keyword, sort and filter the element connections with high coherence, and generate a coherence score table;

[0126] S3: Based on the coherence score table, the keyword connection and interaction information are converted into time series data, the synchronization error is calculated, the query trend is predicted, and the data prediction results are generated;

[0127] S4: Based on the data prediction results, adjust the query parameters, compare the modified parameters with the current state of the information source, and generate optimized query parameters;

[0128] S5: Based on the optimized query parameters, the information retrieval task is re-executed, the keyword priority and the query result order are adjusted, and the optimized query results are generated.

[0129] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. Intelligent query system based on semantic analysis, characterized by: The system comprises: The entity-relationship mapping module collects user query data, performs part-of-speech tagging and syntactic parsing operations on user queries, identifies key elements in query sentences, which are names of people, places, events, or specified objects, builds connection mappings based on the syntactic relationships between key elements, calculates the correlation between each pair of keywords in the mapping, quantifies the intensity of attribute interaction, and generates a keyword relationship mapping table; The coherence analysis module receives the keyword relationship mapping table, calculates the coherence score of each connection according to the key element relationship and attribute interaction information listed in the table, sorts and filters the coherence scores, and generates a coherence score table; The time series matching module converts the keyword connections and attribute interaction information with significant scores in the coherence scoring table and identified from the user query statement into time series data according to the time points when they appear in the text, calculates the synchronization error between the sequences, compares the matching degree of the time series, and predicts the future query trend or information demand change to generate data prediction results; The query parameter optimization module adjusts the query parameters according to the matching degree information of the data prediction results, compares the modified query parameters with the current state of the information source, and reapplies the modified parameters to the information retrieval process to generate optimized query results.

2. The intelligent query system based on semantic analysis according to claim 1, characterized in that: The steps for obtaining the association between each pair of keywords in the calculation map are specifically as follows: Collect user query data, perform word segmentation and part-of-speech tagging, identify key elements in the query, determine the syntactic dependency between elements, and generate key element recognition results; Based on the key element recognition results, analyzing the syntactic relationship between elements, and establishing a connection mapping between key elements to generate a key element connection mapping diagram; According to the key element connection mapping diagram, the formula is used: Calculate the correlation R between keywords A and B AB , considering the number of co-occurrences, grammatical weights, semantic similarity and frequency ratio, to generate keyword relevance analysis results; Among them, C AB represents the number of times keywords A and B appear together in a query, N is the total number of queries in the corpus, and W AB is the weight coefficient based on the syntactic dependency between keywords A and B, F AB is the semantic similarity between keywords A and B, P AB is the frequency ratio of keywords A and B in the entire corpus.

3. The intelligent query system based on semantic analysis according to claim 2 is characterized in that: The steps for obtaining the quantitative attribute interaction strength are specifically as follows: Based on the collected element interaction information, the contextual relationship of key elements is analyzed using the formula: Calculate the attribute interaction strength S between element i and element j ij , generating interaction strength information between elements; Among them, A ij represents the total number of interactions between element i and element j, B ij represents the total number of elements in all interaction events in the corpus, W ij is the interaction weight coefficient, R ij is the context relevance coefficient, which indicates the degree of relevance between element i and element j in the context; According to the interaction intensity information between the elements and the keyword association analysis results, by analyzing the syntactic structure and dependency relationship in the query statement, a relationship network diagram is constructed, all keywords and their interaction information are integrated, and a keyword relationship mapping table is generated.

4. The intelligent query system based on semantic analysis according to claim 3 is characterized in that: The steps for obtaining the coherence score sheet are specifically as follows: Analyzing the keyword relationship in the query sentence according to the keyword relationship mapping table, evaluating the semantic relevance of the keywords in the context by calculating the coherence score between each pair of key elements, and obtaining a keyword coherence analysis result; According to the keyword coherence analysis result, element relationships with significant coherence are screened by analyzing the scores, and a coherence score table is generated.

5. The intelligent query system based on semantic analysis according to claim 4 is characterized in that: The steps for obtaining the matching degree of the comparison time series are specifically as follows: According to the keyword connection and attribute interaction information in the coherence score table, the coherence score is brought in to mark the timestamp according to the appearance time of the keyword, and its priority in the time series is determined to generate time series data; According to the time series data, the formula is adopted: Calculate the synchronization error E between two time series sync ,According to the order of time points in the time series, the differences of keywords at time points are compared, and the matching between sequences is analyzed to generate time series matching results; Among them, X t represents the value of the time series X at the tth time point, Y t represents the value of the time series Y at the tth time point, L X and L Y Respectively represent the length of time series X and Y, T max It represents the maximum time point value that occurs in the two time series, and n represents the total number of time points.

6. The intelligent query system based on semantic analysis according to claim 5, characterized in that: The steps for obtaining the predicted future query trend or information demand change are specifically as follows: According to the time series matching results, combined with the semantic relevance and keyword interaction frequency information, the formula is adopted: Calculate the predicted time point T pred , generate prediction time point information; Among them, T curr is the current time point, D rel is the semantic relevance, n is the total number of time points, F int is the interaction frequency; According to the predicted time point information, the data segments most relevant to the query trend are identified, and the weights and rankings are adjusted according to the semantic similarity of the keywords to generate data prediction results.

7. The intelligent query system based on semantic analysis according to claim 6, characterized in that: The steps for obtaining the optimized query results are specifically as follows: According to the data prediction results, the keyword weight is evaluated, and the importance of the keyword is corrected according to the synchronization error, the query parameter is modified, and an adjusted query parameter configuration is generated; According to the adjusted query parameter configuration, the priority and weight of the keyword are recalculated, the query task is executed, and the query result sequence is adjusted according to the keyword weight to obtain an optimized query result.

8. An intelligent query method based on semantic analysis, characterized in that: The intelligent query system based on semantic analysis according to any one of claims 1 to 7 comprises the following steps: Based on user query data, perform part-of-speech tagging and syntactic analysis, identify key elements, build connection mappings between key elements, calculate the correlation between keywords, and generate a keyword relationship mapping table; Based on the keyword relationship mapping table, the coherence score between each keyword is calculated, and the element connections with high coherence are sorted and screened to generate a coherence score table; Based on the coherence score table, the keyword connection and interaction information are converted into time series data, the synchronization error is calculated, the query trend is predicted, and the data prediction result is generated; Based on the data prediction results, adjusting the query parameters, comparing the modified parameters with the current state of the information source, and generating optimized query parameters; Based on the optimized query parameters, the information retrieval task is re-executed, the keyword priority and the query result sequence are adjusted, and the optimized query result is generated.

Images