Archives retrieval method and system based on big data analysis

Abstract

The application discloses an archive retrieval method and system based on big data analysis and belongs to the technical field of information retrieval. The method comprises the following steps: initializing a retrieval session; receiving a first ambiguous query in the retrieval session to obtain a first ambiguous query analysis result; establishing an initial logical context; obtaining an analyzed query when it is detected that an ambiguous mode of a subsequent ambiguous query matches a logical feature recorded in a logical preference template; monitoring user behavior signals in the retrieval session and dynamically updating logical features in the logical preference template according to the user behavior signals. The application solves the problems of uncertain query analysis, chaotic result sorting and forced distortion of user retrieval behavior caused by fixed priority rules in the prior art by establishing a session-level logical preference template and dynamically maintaining user query habit features.

Description

Technical Field

[0001] This invention relates to the field of information retrieval technology, and in particular to a method and system for archival retrieval based on big data analysis. Background Technology

[0002] With the deepening of information technology development, various institutions have accumulated massive amounts of electronic archival data. How to quickly and accurately retrieve the required information from this vast amount of data has become crucial for improving work efficiency. The introduction of big data analytics technology provides new technical support for the intelligent development of archival retrieval systems.

[0003] Existing document retrieval systems typically achieve querying of structured and unstructured documents by constructing inverted indexes and using Boolean logic models. When parsing complex queries containing multiple Boolean operators, these systems mainly rely on predefined fixed priority rules, lacking the ability to adapt to users' personalized query habits.

[0004] In existing technologies, the query parsing module's mechanism for handling operator priority is rather rigid, resulting in insufficient accuracy in interpreting query intent. Specifically, when users input complex queries using a mix of AND, OR, and other operators, the system cannot intelligently determine the operator priority based on context or user habits, leading to different parsing results for the same query in different scenarios. This parsing uncertainty directly triggers a chain reaction of problems related to the disordered sorting logic of search results, making it difficult to prioritize highly relevant files. Furthermore, this user experience deficiency forces users to repeatedly add parentheses or split queries to clarify their intent, causing a deeper problem of distorting natural user query behavior. This not only increases the user's search burden but also distorts the query data collected by the system, ultimately creating a negative cycle that hinders system optimization and upgrades. Summary of the Invention

[0005] This application provides a file retrieval method and system based on big data analysis, which solves the problems of chaotic retrieval results and distorted user query behavior caused by unclear query priority parsing rules in the prior art. It achieves the technical effect of intelligently eliminating query ambiguity and ensuring the accuracy and consistency of retrieval results by dynamically learning the logical preferences in user sessions.

[0006] This application provides a file retrieval method based on big data analysis, including: initializing a retrieval session and creating a session-unique logical preference template for the retrieval session; Receive the first ambiguous query in the retrieval session, parse the first ambiguous query using the retrieval baseline parsing rules, and obtain the parsing result of the first ambiguous query; Extract logical features from the parsing result of the first ambiguous query to fill the logical preference template and establish the initial logical context; Receive subsequent ambiguous queries in the retrieval session. When the ambiguous pattern of the subsequent ambiguous query is detected to match the logical features recorded in the logical preference template, apply the logical preferences corresponding to the logical features to parse the subsequent ambiguous query and obtain the parsed query. Monitor user behavior signals in the retrieval session and dynamically update the logical features in the logical preference template based on the user behavior signals.

[0007] Furthermore, the steps of initializing the retrieval session and creating a session-unique logical preference template for the retrieval session include: When a user initiates a query request for the first time without authentication, the retrieval session process is started, and a retrieval session instance is generated by assigning a unique identifier. Create a logical preference template structure in the retrieval session instance. The logical preference template structure contains a logical feature storage area and a preference mapping area. Initialize the logical feature storage area to an empty set and set the preference mapping area to the default empty mapping; Load the list of standard logical feature types from the preset logical feature type library, including time logical features, spatial logical features, and entity logical features; Allocate memory space for the logical preference template and establish an access control mechanism for the template, allowing only the current retrieval session instance to read and write the template content.

[0008] Furthermore, the step of parsing the first ambiguous query using retrieval benchmark parsing rules to obtain the parsing result of the first ambiguous query includes: The received query string is lexically analyzed and decomposed into operand units and operator units; If a query structure is found to contain at least two Boolean operators of different types and there is no explicit bracket grouping relationship between these operators, it is considered an ambiguous query. Activate the benchmark parsing rule base, which predefines the default priority order of Boolean operators; The query units are logically grouped according to the default priority order in a left-to-right scanning order to generate an abstract syntax tree; Based on the constructed abstract syntax tree, a standardized query expression is generated, completing the initial ambiguity resolution process.

[0009] Furthermore, the step of extracting logical features from the parsing result of the first ambiguous query to fill the logical preference template and establish the initial logical context includes: Analyze the abstract syntax tree structure in the first generated ambiguous query parsing result and identify the key nodes in the tree structure that reflect priority decisions; Extract operator type combination features from the key nodes; Extract priority decision features, associate and encapsulate the extracted operator type combination features with the priority decision features to form a complete logical rule record; A strength index value is initialized for the complete logical rule record, the strength index value representing the confidence level of this logical rule in the current session; The encapsulated logical rules are recorded in the logical rule record table of the logical preference template, completing the initial filling process of the template and establishing the session-specific initial logical context.

[0010] Furthermore, the step of detecting that the ambiguity pattern of the subsequent ambiguous query matches the logical features recorded in the logical preference template includes: Before performing real-time parsing analysis on subsequent received query requests, identify the types of Boolean operators contained in the query string and their distribution positions in the string; Based on the identified Boolean operator types and their distribution locations, generate an ambiguous pattern descriptor for the current query; Compare the ambiguous pattern descriptor of the current query with the existing logical rule records stored in the logical preference template item by item; The comparison process first checks whether the combination of operator types is consistent; When the operator type combination features are consistent, further verify whether the relative positional relationship of the operators in the query string matches the positional relationship pattern in the template record; Calculate the matching score between the current query ambiguity pattern descriptor and each logical rule record in the template. When the matching score of at least one logical rule record exceeds the preset matching threshold, it is determined that a match has been detected. The logical rule record with the highest matching score is selected as the guiding principle for subsequent query parsing.

[0011] Furthermore, the step of dynamically updating the logical features in the logical preference template based on the user behavior signals includes: Continuously monitor user interaction behavior during retrieval sessions and capture two types of user behavior signals; The first type of user behavior signal is the user's explicit modification of the query structure. When it is detected that the user has manually modified the parsed query and added explicit bracket grouping to the modified query, the logical preference template update process is immediately triggered. The second type is the session duration signal, which records the interval between the last update time of the logical rule record and the current time; For the first type of behavior signal, analyze the user's modified query structure, extract the new priority relationship, find the logical rule records that conflict with it in the logical preference template, reduce the intensity index value of these logical rule records, and create logical rule records corresponding to the new priority relationship. For the second type of behavioral signal, all logical rule records in the logical preference template are scanned periodically, and the intensity index value of each record is reduced according to the preset decay function based on the time interval. When the strength index value of a logical rule record is lower than the set failure threshold, the record is removed from the logical rule record table.

[0012] Furthermore, the steps for calculating the matching score between the current query ambiguity pattern descriptor and each logical rule record in the template include: Get the set of operator types for the current query And operator position sequence Simultaneously, extract the set of operator types of the records from the logical rule records. And operator position sequence ; The type matching degree is calculated by comparing the set of operator types, the structural similarity is calculated by comparing the position sequence of operators, and the final matching degree score is calculated by combining the strength index in the logical rule record. Match score The calculation formula is: ; In the formula, , , These are the weighting coefficients. The position attenuation coefficient, For the normalized value of the intensity index in the logical rule record, It is a natural constant.

[0013] This application provides an archival retrieval system based on big data analysis, used to implement an archival retrieval method based on big data analysis, including: The module includes a session initialization module, a first query parsing module, a feature-filled template module, a subsequent query matching module, and a template dynamic update module. The session initialization module is used to initialize a retrieval session and create a unique logical preference template for the retrieval session. The first query parsing module is used to receive the first ambiguous query in the retrieval session, parse the first ambiguous query using the retrieval benchmark parsing rules, and obtain the first ambiguous query parsing result; The feature filling template module is used to extract logical features from the parsing result of the first ambiguous query and fill the logical preference template to establish an initial logical context. The subsequent query matching module is used to receive subsequent ambiguous queries in the retrieval session. When the ambiguous pattern of the subsequent ambiguous query is detected to match the logical features recorded in the logical preference template, the logical preference corresponding to the logical feature is applied to parse the subsequent ambiguous query to obtain the parsed query. The template is dynamically updated to monitor user behavior signals in the retrieval session and to dynamically update the logical features in the logical preference template based on the user behavior signals.

[0014] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: By capturing users' logical preferences during conversations, the system can intelligently adapt to individual query habits and automatically eliminate priority ambiguities in complex queries, thereby significantly improving the accuracy and consistency of query parsing. Furthermore, building upon improved query parsing accuracy, the ability to accurately interpret query intent based on learned logical preferences allows the retrieval execution phase to match files according to a logical structure that aligns with the user's actual needs. This fundamentally avoids the problem of result sorting chaos caused by parsing errors, ensuring that highly relevant files are presented to the user first.

[0015] Furthermore, by continuously monitoring user behavior signals and dynamically updating logical preference templates, the system can adapt to changes in user query habits in a timely manner, forming a virtuous cycle of learning. This effectively prevents the behavioral distortion of users being forced to adjust their query methods in order to obtain the expected results, and allows for continuous optimization based on real query behavior data. Attached Figure Description

[0016] Figure 1 A flowchart of a file retrieval method based on big data analysis provided in this application embodiment; Figure 2 This is a schematic diagram of the structure of the archive retrieval system based on big data analysis provided in an embodiment of this application. Detailed Implementation

[0017] This application provides a file retrieval method and system based on big data analysis, which solves the problems of unclear query priority parsing rules leading to chaotic search result sorting and distorted user query behavior in the prior art. By initializing logical preference templates for the retrieval session, extracting and applying user logical features to parse ambiguous queries, and dynamically updating template content, it achieves accurate parsing of query intent and consistent presentation of search results, thereby improving retrieval efficiency and user experience.

[0018] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0019] like Figure 1 The diagram shown is a flowchart of a document retrieval method based on big data analysis provided in this application embodiment. The method is applied to a document retrieval system based on big data analysis and includes the following steps: initializing a retrieval session and creating a session-unique logical preference template for the retrieval session; Receive the first ambiguous query in the retrieval session, parse the first ambiguous query using the retrieval baseline parsing rules, and obtain the parsing result of the first ambiguous query; Extract logical features from the parsing result of the first ambiguous query to fill the logical preference template and establish the initial logical context; Receive subsequent ambiguous queries in the retrieval session. When the ambiguous pattern of the subsequent ambiguous query is detected to match the logical features recorded in the logical preference template, apply the logical preferences corresponding to the logical features to parse the subsequent ambiguous query and obtain the parsed query. Monitor user behavior signals in the retrieval session and dynamically update the logical features in the logical preference template based on the user behavior signals.

[0020] Furthermore, the steps of initializing the retrieval session and creating a session-unique logical preference template for the retrieval session include: When a user initiates a query request for the first time without authentication, the retrieval session process is started, and a retrieval session instance is generated by assigning a unique identifier. Create a logical preference template structure in the retrieval session instance. The logical preference template structure contains a logical feature storage area and a preference mapping area. Initialize the logical feature storage area to an empty set and set the preference mapping area to the default empty mapping; Load the list of standard logical feature types from the preset logical feature type library, including time logical features, spatial logical features, and entity logical features; Allocate memory space for logical preference templates to ensure that the templates are unique throughout the lifetime of the retrieval session; Establish an access control mechanism for the template, allowing only the current retrieval session instance to read and write the template content.

[0021] Furthermore, the step of parsing the first ambiguous query using retrieval benchmark parsing rules to obtain the parsing result of the first ambiguous query includes: The received query string is lexically analyzed and decomposed into operand units and operator units; If a query structure is found to contain at least two Boolean operators of different types and there is no explicit bracket grouping relationship between these operators, it is considered an ambiguous query. Activate the baseline parsing rule base, which predefines the default priority order of Boolean operators, namely, the NOT operator has the highest parsing priority, the AND operator has the second highest parsing priority, and the OR operator has the lowest parsing priority; Following the scanning order from left to right, the query units are logically grouped based on the default priority order to generate an abstract syntax tree with a clear hierarchical structure; When a query contains both AND and OR operators and there are no parentheses for explicit grouping, the AND operator and its adjacent operands are combined into a logical unit, and then this logical unit is combined with the OR operator. Based on the constructed abstract syntax tree, a standardized query expression is generated, completing the initial ambiguity resolution process.

[0022] Furthermore, the step of extracting logical features from the parsing result of the first ambiguous query to fill the logical preference template and establish the initial logical context includes: Analyze the abstract syntax tree structure in the first generated ambiguous query parsing result and identify the key nodes in the tree structure that reflect priority decisions; Determining whether a node in an abstract syntax tree (AST) is a key node primarily relies on its position within the tree structure and its decisive role in the final query's logical grouping and order of operations. Specifically, key nodes typically refer to those connecting operators of different precedence (such as AND and OR), or branches where the subtree structure undergoes significant changes; the parsing method of these nodes directly alters the semantics of the entire query statement. In the tree structure, branches and intersections are also considered key nodes due to their significant impact on the topology. Furthermore, during classification or discrimination, non-terminal nodes (nodes with child nodes) are generally more critical than terminal nodes because they perform logical discrimination functions. The system identifies these key nodes by analyzing their attributes and relative positions within the tree.

[0023] Operator type combination features are extracted from the key nodes. These features record the types of Boolean operators involved in the parsing process and their relative positions in the syntax tree. Extract priority decision features, which explicitly record the final priority order selected among multiple potentially ambiguous parsing paths; The extracted operator type combination features are associated and encapsulated with priority decision features to form a complete logical rule record; A strength index value is initialized for the complete logical rule record, the strength index value representing the confidence level of this logical rule in the current session; The encapsulated logical rules are recorded in the logical rule record table of the logical preference template, completing the initial filling process of the template and establishing the session-specific initial logical context.

[0024] Furthermore, the step of detecting that the ambiguity pattern of the subsequent ambiguous query matches the logical features recorded in the logical preference template includes: Before performing real-time parsing analysis on subsequent received query requests, identify the types of Boolean operators contained in the query string and their distribution positions in the string; Based on the identified Boolean operator types and their distribution locations, generate an ambiguous pattern descriptor for the current query; Compare the ambiguous pattern descriptor of the current query with the existing logical rule records stored in the logical preference template item by item; The comparison process first checks whether the operator type combination features are consistent, that is, whether the types of operators contained in the current query are exactly the same as the types of operators in a certain record in the template; When the operator type combination features are consistent, further verify whether the relative positional relationship of the operators in the query string matches the positional relationship pattern in the template record; Calculate the matching score between the current query ambiguity pattern descriptor and each logical rule record in the template. When the matching score of at least one logical rule record exceeds the preset matching threshold, it is determined that a match has been detected. The logical rule record with the highest matching score is selected as the guiding principle for subsequent query parsing.

[0025] Furthermore, the step of dynamically updating the logical features in the logical preference template based on the user behavior signals includes: Continuously monitor user interaction behavior during retrieval sessions and capture two types of user behavior signals; The first type of user behavior signal is the user's explicit modification of the query structure. When it is detected that the user has manually modified the parsed query and added explicit bracket grouping to the modified query, the logical preference template update process is immediately triggered. The second type is the session duration signal, which records the interval between the last update time of the logical rule record and the current time; For the first type of behavior signal, analyze the user's modified query structure, extract the new priority relationship, find the logical rule records that conflict with it in the logical preference template, reduce the intensity index value of these logical rule records, and create logical rule records corresponding to the new priority relationship. For the second type of behavioral signal, all logical rule records in the logical preference template are scanned periodically, and the intensity index value of each record is reduced according to the preset decay function based on the time interval. The preset decay function is: ; In the formula, The intensity index value at the current time point is a non-negative real number, representing the confidence level of the logical rule recorded at time t. The initial strength index value is a non-negative real number representing the initial confidence level of the logical rule record at the time of its creation or last update. This represents the attenuation constant, a positive real number that controls the attenuation rate of the intensity index value. Its value is preset by the system, determining the rate of attenuation. Representing the time interval, it is a non-negative real number representing the time difference between the last update timestamp of the logical rule record and the current system timestamp, usually calculated in seconds. It represents the base of the natural logarithm.

[0026] When the strength index value of a certain logical rule record is lower than the set failure threshold, the record is removed from the logical rule record table; the above mechanism realizes the dynamic maintenance of logical features in the logical preference template.

[0027] Furthermore, the steps for calculating the matching score between the current query ambiguity pattern descriptor and each logical rule record in the template include: Get the set of operator types for the current query AND operator position sequence Simultaneously, extract the set of operator types of the records from the logical rule records. AND operator position sequence ; The type matching degree is calculated by comparing the set of operator types, the structural similarity is calculated by comparing the position sequence of operators, and the final matching degree score is calculated by combining the strength index in the logical rule record. Match score The calculation formula is: ; In the formula, , , The weighting coefficients are satisfied. , The position attenuation coefficient, For the normalized value of the intensity index in the logical rule record, The set of operator types is obtained by scanning the query string to identify AND, OR, and NOT operators; the operator position sequence is obtained by recording and normalizing the relative position indices of each operator in the query string; the normalized value of the intensity index is obtained by linearly transforming the recorded intensity index to the [0,1] interval; the position difference norm is obtained by calculating the Euclidean distance between two position sequences; and the weighting coefficient is... , , and position attenuation coefficient Fixed values ​​are obtained through training with historical session data.

[0028] like Figure 2 The diagram shown is a structural schematic of the archive retrieval system based on big data analysis provided in this application embodiment. The archive retrieval system based on big data analysis provided in this application embodiment includes: a session initialization module, a first query parsing module, a feature filling template module, a subsequent query matching module, and a template dynamic update module. The session initialization module is used to initialize a retrieval session and create a unique logical preference template for the retrieval session. The first query parsing module is used to receive the first ambiguous query in the retrieval session, parse the first ambiguous query using the retrieval benchmark parsing rules, and obtain the first ambiguous query parsing result; The feature filling template module is used to extract logical features from the parsing result of the first ambiguous query and fill the logical preference template to establish an initial logical context. The subsequent query matching module is used to receive subsequent ambiguous queries in the retrieval session. When the ambiguous pattern of the subsequent ambiguous query is detected to match the logical features recorded in the logical preference template, the logical preference corresponding to the logical feature is applied to parse the subsequent ambiguous query to obtain the parsed query. The template is dynamically updated to monitor user behavior signals in the retrieval session and to dynamically update the logical features in the logical preference template based on the user behavior signals.

[0029] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0030] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0031] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0032] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0033] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0034] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for retrieving archives based on big data analysis, characterized in that: Includes the following steps: Initialize the retrieval session and create a session-unique logical preference template for the retrieval session; The steps for creating a session-unique logical preference template for the retrieval session include: When a user initiates a query request for the first time without authentication, the retrieval session process is started, and a retrieval session instance is generated by assigning a unique identifier. Create a logical preference template structure in the retrieval session instance. The logical preference template structure contains a logical feature storage area and a preference mapping area. Initialize the logical feature storage area to an empty set and set the preference mapping area to the default empty mapping; Load the list of standard logical feature types from the preset logical feature type library, including time logical features, spatial logical features, and entity logical features; Allocate memory space for the logical preference template and establish an access control mechanism for the template, allowing only the current retrieval session instance to read and write the template content; Receive the first ambiguous query in the retrieval session, parse the first ambiguous query using the retrieval baseline parsing rules, and obtain the parsing result of the first ambiguous query; The steps to obtain the first ambiguous query resolution result include: The received query string is lexically analyzed and decomposed into operand units and operator units; If a query structure is found to contain at least two Boolean operators of different types and there is no explicit bracket grouping relationship between these operators, it is considered an ambiguous query. Activate the benchmark parsing rule base, which predefines the default priority order of Boolean operators; The query units are logically grouped according to the default priority order in a left-to-right scanning order to generate an abstract syntax tree; Based on the constructed abstract syntax tree, a standardized query expression is generated, completing the initial ambiguity resolution process; Extract logical features from the parsing result of the first ambiguous query to fill the logical preference template and establish the initial logical context; Receive subsequent ambiguous queries in the retrieval session. When the ambiguous pattern of the subsequent ambiguous query is detected to match the logical features recorded in the logical preference template, apply the logical preferences corresponding to the logical features to parse the subsequent ambiguous query and obtain the parsed query. Monitor user behavior signals in the retrieval session and dynamically update the logical features in the logical preference template based on the user behavior signals; The steps of dynamically updating the logical features in the logical preference template based on the user behavior signals include: Continuously monitor user interaction behavior during retrieval sessions and capture two types of user behavior signals; The first type of user behavior signal is the user's explicit modification of the query structure. When it is detected that the user has manually modified the parsed query and added explicit bracket grouping to the modified query, the logical preference template update process is immediately triggered. The second type is the session duration signal, which records the interval between the last update time of the logical rule record and the current time; For the first type of behavior signal, analyze the user's modified query structure, extract the new priority relationship, find the logical rule records that conflict with it in the logical preference template, reduce the intensity index value of these logical rule records, and create logical rule records corresponding to the new priority relationship. For the second type of behavioral signal, all logical rule records in the logical preference template are scanned periodically, and the intensity index value of each record is reduced according to the preset decay function based on the time interval. When the strength index value of a logical rule record is lower than the set failure threshold, the record is removed from the logical rule record table.

2. The archive retrieval method based on big data analysis as described in claim 1, characterized in that, The steps of extracting logical features from the first ambiguous query parsing result to fill the logical preference template and establish the initial logical context include: Analyze the abstract syntax tree structure in the first generated ambiguous query parsing result and identify the key nodes in the tree structure that reflect priority decisions; Extract operator type combination features from the key nodes; Extract priority decision features, associate and encapsulate the extracted operator type combination features with the priority decision features to form a complete logical rule record; A strength index value is initialized for the complete logical rule record, the strength index value representing the confidence level of this logical rule in the current session; The encapsulated logical rules are recorded in the logical rule record table of the logical preference template, completing the initial filling process of the template and establishing the session-specific initial logical context.

3. The archive retrieval method based on big data analysis as described in claim 1, characterized in that, The step of detecting that the ambiguous pattern of the subsequent ambiguous query matches the logical features recorded in the logical preference template includes: Before performing real-time parsing analysis on subsequent received query requests, identify the types of Boolean operators contained in the query string and their distribution positions in the string; Based on the identified Boolean operator types and their distribution locations, generate an ambiguous pattern descriptor for the current query; Compare the ambiguous pattern descriptor of the current query with the existing logical rule records stored in the logical preference template item by item; The comparison process first checks whether the combination of operator types is consistent; When the operator type combination features are consistent, further verify whether the relative positional relationship of the operators in the query string matches the positional relationship pattern in the template record; Calculate the matching score between the current query ambiguity pattern descriptor and each logical rule record in the template. When the matching score of at least one logical rule record exceeds the preset matching threshold, it is determined that a match has been detected. The logical rule record with the highest matching score is selected as the guiding principle for subsequent query parsing.

4. The archive retrieval method based on big data analysis as described in claim 3, characterized in that, The steps for calculating the match score between the current query ambiguity pattern descriptor and each logical rule record in the template include: Get the set of operator types for the current query AND operator position sequence Simultaneously, extract the set of operator types of the records from the logical rule records. AND operator position sequence ; The type matching degree is calculated by comparing the set of operator types, the structural similarity is calculated by comparing the position sequence of operators, and the final matching degree score is calculated by combining the strength index in the logical rule record. Match score The calculation formula is: ； In the formula, , , These are the weighting coefficients. The position attenuation coefficient, For the normalized value of the intensity index in the logical rule record, It is a natural constant.

5. A big data-based archival retrieval system, used to implement the big data-based archival retrieval method according to any one of claims 1-4, characterized in that, include: The module includes a session initialization module, a first query parsing module, a feature-filled template module, a subsequent query matching module, and a template dynamic update module. The session initialization module is used to initialize a retrieval session and create a unique logical preference template for the retrieval session. The steps for creating a session-unique logical preference template for the retrieval session include: When a user initiates a query request for the first time without authentication, the retrieval session process is started, and a retrieval session instance is generated by assigning a unique identifier. Create a logical preference template structure in the retrieval session instance. The logical preference template structure contains a logical feature storage area and a preference mapping area. Initialize the logical feature storage area to an empty set and set the preference mapping area to the default empty mapping; Load the list of standard logical feature types from the preset logical feature type library, including time logical features, spatial logical features, and entity logical features; Allocate memory space for the logical preference template and establish an access control mechanism for the template, allowing only the current retrieval session instance to read and write the template content; The first query parsing module is used to receive the first ambiguous query in the retrieval session, parse the first ambiguous query using the retrieval benchmark parsing rules, and obtain the first ambiguous query parsing result; The steps to obtain the first ambiguous query resolution result include: The received query string is lexically analyzed and decomposed into operand units and operator units; If a query structure is found to contain at least two Boolean operators of different types and there is no explicit bracket grouping relationship between these operators, it is considered an ambiguous query. Activate the benchmark parsing rule base, which predefines the default priority order of Boolean operators; The query units are logically grouped according to the default priority order in a left-to-right scanning order to generate an abstract syntax tree; Based on the constructed abstract syntax tree, a standardized query expression is generated, completing the initial ambiguity resolution process; The feature filling template module is used to extract logical features from the parsing result of the first ambiguous query and fill the logical preference template to establish an initial logical context. The subsequent query matching module is used to receive subsequent ambiguous queries in the retrieval session. When the ambiguous pattern of the subsequent ambiguous query is detected to match the logical features recorded in the logical preference template, the logical preference corresponding to the logical feature is applied to parse the subsequent ambiguous query to obtain the parsed query. The template is dynamically updated to monitor user behavior signals in the retrieval session and to dynamically update the logical features in the logical preference template based on the user behavior signals. The steps of dynamically updating the logical features in the logical preference template based on the user behavior signals include: Continuously monitor user interaction behavior during retrieval sessions and capture two types of user behavior signals; The first type of user behavior signal is the user's explicit modification of the query structure. When it is detected that the user has manually modified the parsed query and added explicit bracket grouping to the modified query, the logical preference template update process is immediately triggered. The second type is the session duration signal, which records the interval between the last update time of the logical rule record and the current time; For the first type of behavior signal, analyze the user's modified query structure, extract the new priority relationship, find the logical rule records that conflict with it in the logical preference template, reduce the intensity index value of these logical rule records, and create logical rule records corresponding to the new priority relationship. For the second type of behavioral signal, all logical rule records in the logical preference template are scanned periodically, and the intensity index value of each record is reduced according to the preset decay function based on the time interval. When the strength index value of a logical rule record is lower than the set failure threshold, the record is removed from the logical rule record table.

Images