Data query analysis method and system based on LLM and semantic model

Abstract

The invention discloses a data query analysis method and system based on an LLM and a semantic model, and the method comprises the steps: carrying out the analysis of a target query statement inputted by a user, so as to obtain a retrieval data set corresponding to the target query statement; retrieving similar questions in a vector database based on the target query statement to obtain the similar questions; analyzing the retrieval data set and the similar problem by using a large language model (LLM) to obtain a semantic SQL statement; and analyzing the semantic SQL statement into a query SQL statement by utilizing a semantic model, and returning a query result to a user based on the SQL query statement. According to the method, the result obtained by performing data query analysis in the dialogue mode is more accurate, the cost is lower through the mode of combining the semantic model and the LLM, and due to the fact that the complexity of generating the large model SQL is reduced, the generation of the data query SQL can be achieved only through the LLM with small parameters, and the system application cost is greatly reduced.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence and data processing technology, and more specifically, to a data query analysis method and system based on LLM and semantic model. Background Art

[0002] LLM (Large Language Model) is a machine learning model that can perform various Natural Language Processing (NLP) tasks, such as creating content, translating text, or parsing sentences. The key feature of LLM is that it can understand and generate natural language. To achieve this goal, LLM requires a large amount of data and computing resources for training.

[0003] In traditional enterprises, BI (Business Intelligence) systems are used to effectively integrate enterprise data and quickly generate reports for decision-making. However, in actual work, the problems to be analyzed may not be fixed, and answers cannot usually be obtained directly from the current reports. Instead, business personnel, data analysts, and technical personnel need to repeatedly edit reports and add more indicators and dimensions to gain business insights into the problems. The communication process is cumbersome and inefficient.

[0004] At present, AI GC's conversational BI products in Business Intelligence (BI), such as ChatBI, can recognize and understand users' natural language questions and interact with users. However, the current conversational BI products' ability to understand users' questions is still very limited. When conversational BI products understand natural language, there are problems with converting structured query language SQL and failing to accurately generate SQL to implement complex logic, which makes conversational BI products still at a relatively low level of intelligence.

[0005] To address the above problems, no effective solution has been proposed yet.

[0006] Therefore, a data query and analysis method based on LLM and semantic model is needed. Summary of the invention

[0007] The present invention proposes a data query analysis method and system based on LLM and semantic model to solve the problem of how to efficiently implement conversational data query.

[0008] In order to solve the above problem, according to one aspect of the present invention, a data query analysis method based on LLM and semantic model is provided, the method comprising:

[0009] Parsing the target query statement input by the user to obtain a retrieval data set corresponding to the target query statement;

[0010] Retrieving similar questions in a vector database based on the target query statement to obtain similar questions;

[0011] Analyze the retrieval data set and similar questions using a large language model (LLM) to obtain semantic SQL statements;

[0012] The semantic SQL statement is parsed into a query SQL statement using a semantic model, and a query result is returned to the user based on the SQL query statement.

[0013] Preferably, the method further comprises:

[0014] Based on the target query statement, similar questions are searched in a vector database by using a cosine similarity or Euclidean distance method to obtain similar questions.

[0015] Preferably, the method further comprises:

[0016] Divide the subject domain according to the subject application, and establish the data source to be connected according to the divided subject domain;

[0017] Configure the model for the database table in the selected data source and define the fields as atomic indicators and dimension dates;

[0018] Create new indicators and dimensions to be analyzed based on requirements, and define tags to associate with related fields;

[0019] The indicators, dimensions, and labels defined by the subject domain are added to the dataset so that they can be passed to the LLM model as a retrieval dataset when making queries.

[0020] Preferably, the method further comprises:

[0021] The pre-set questions and SQL pair knowledge obtained after vectorization using the Text2vec method are stored in the vector database, and the vector database is enriched according to user queries.

[0022] Preferably, the method further comprises:

[0023] When the data to be analyzed will affect the business database and the analysis efficiency in the business database is not high, the data to be analyzed is extracted to the analysis database for analysis;

[0024] The query results are automatically displayed in a chart, and a multi-dimensional drill-down query is performed based on the query results, and the multi-dimensional drill-down query results are displayed.

[0025] According to another aspect of the present invention, a data query analysis system based on LLM and semantic model is provided, the system comprising:

[0026] A target query statement parsing unit, used to parse the target query statement input by the user to obtain a retrieval data set corresponding to the target query statement;

[0027] A similar question determining unit, configured to retrieve similar questions in a vector database based on the target query statement to obtain similar questions;

[0028] A semantic SQL statement acquisition unit, used for analyzing the retrieval data set and similar questions using a large language model LLM to acquire a semantic SQL statement;

[0029] The query result acquisition unit is used to parse the semantic SQL statement into a query SQL statement by using a semantic model, and return the query result to the user based on the SQL query statement.

[0030] Preferably, the formal problem determination unit further comprises:

[0031] Based on the target query statement, similar questions are searched in a vector database through a cosine similarity or Euclidean distance system to obtain similar questions.

[0032] Preferably, the system further comprises: a configuration unit, configured to:

[0033] Divide the subject domain according to the subject application, and establish the data source to be connected according to the divided subject domain;

[0034] Configure the model for the database table in the selected data source and define the fields as atomic indicators and dimension dates;

[0035] Create new indicators and dimensions to be analyzed based on requirements, and define tags to associate with related fields;

[0036] The indicators, dimensions, and labels defined by the subject domain are added to the dataset so that they can be passed to the LLM model as a retrieval dataset when making queries.

[0037] Preferably, the system further comprises:

[0038] The vector database determination unit is used to store the pre-set question and SQL pair knowledge obtained after vectorization using the Text2vec method in the vector database, and enrich the vector database according to the user query.

[0039] Preferably, the system further comprises:

[0040] An analysis database determination unit, used for extracting the data to be analyzed into the analysis database for analysis when the data to be analyzed will affect the business database and the analysis efficiency in the business database is not high;

[0041] A display unit, used to automatically display the query results in a chart; and to display multi-dimensional drill-down query results;

[0042] The multi-dimensional drill-down query unit is used to perform a multi-dimensional drill-down query based on the query result, and obtain the multi-dimensional drill-down query result.

[0043] The present invention provides a data query analysis method and system based on LLM and semantic model, including: parsing a target query statement input by a user to obtain a retrieval data set corresponding to the target query statement; retrieving similar questions in a vector database based on the target query statement to obtain similar questions; analyzing the retrieval data set and similar questions using a large language model LLM to obtain a semantic SQL statement; parsing the semantic SQL statement into a query SQL statement using a semantic model, and returning the query result to the user based on the SQL query statement. The present invention uses a dialog-based data query analysis method to obtain more accurate results, and the combination of a semantic model and LLM has lower costs. Since the complexity of large model SQL generation is reduced, only a small parameter LLM is required to achieve data query SQL generation, which greatly reduces the system application cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] A more complete understanding of exemplary embodiments of the present invention may be obtained by referring to the following drawings:

[0045] Figure 1 is a flow chart of a data query analysis method 100 based on LLM and semantic model according to an embodiment of the present invention;

[0046] Figure 2 is an overall architecture diagram of a data query function according to an embodiment of the present invention;

[0047] Figure 3 A flowchart of a conversational data query according to an embodiment of the present invention;

[0048] Figure 4 Schematic diagram of the structure of a data query and analysis system 400 based on LLM and semantic model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0049] Reference is now made to the accompanying drawings to describe exemplary embodiments of the present invention. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. These embodiments are provided to disclose the present invention in detail and completely, and to fully convey the scope of the present invention to those skilled in the art. The terms in the exemplary embodiments shown in the drawings are not intended to limit the present invention. In the drawings, the same units / components are denoted by the same reference numerals.

[0050] Unless otherwise specified, the terms (including scientific and technical terms) used herein have the ordinary meaning understood by those skilled in the art. In addition, it can be understood that the terms defined in the commonly used dictionary should be understood to have a meaning consistent with the context of their related fields, and should not be understood as idealized or overly formal meanings.

[0051] The present invention provides a data query and analysis method and system based on an LLM and a semantic model. By combining the LLM large model and the semantic model, semantic definitions such as index dimensions are performed through visual operations, and the indexes are associated with SQL connections or formulas to generate a dataset related to the topic. The user asks data query questions in a conversational manner. The LLM parses the input statement, retrieves the existing Q&A knowledge base through vector retrieval, and transmits the similar Q&A pairs and the dataset related to the topic to the LLM. The LLM generates semantic SQL, locally parses the semantics to generate an actual SQL query statement, and executes the SQL query, and returns the query result to the user. The present invention incorporates data semantics into the prompt words, enabling the LLM to better understand semantics to reduce hallucinations. Placing advanced SQL syntax (such as joins, formulas, etc.) at the semantic layer reduces the complexity of query generation and improves query accuracy.

[0052] Figure 1 FIG. 100 is a flowchart of a data query and analysis method 100 based on an LLM and a semantic model according to an embodiment of the present invention. The present invention provides a data query and analysis method based on an LLM and a semantic model. The result obtained by performing data query and analysis in a conversational manner is more accurate, and the cost is lower by combining the semantic model and the LLM. Since the complexity of generating SQL by the large model is reduced, only an LLM with small parameters is required to implement the generation of data query SQL, greatly reducing the system application cost. The data query and analysis method 100 based on an LLM and a semantic model provided by the embodiment of the present invention starts from step 101. In step 101, the target query statement input by the user is parsed to obtain a retrieval dataset corresponding to the target query statement.

[0053] In step 102, similar questions are retrieved in the vector database based on the target query statement to obtain similar questions.

[0054] Preferably, the method further comprises:

[0055] Based on the target query statement, similar questions are searched in a vector database by using a cosine similarity or Euclidean distance method to obtain similar questions.

[0056] In step 103, the search data set and similar questions are analyzed using a large language model (LLM) to obtain semantic SQL statements.

[0057] In step 104, the semantic SQL statement is parsed into a query SQL statement using a semantic model, and a query result is returned to the user based on the SQL query statement.

[0058] Preferably, the method further comprises:

[0059] Divide the subject domain according to the subject application, and establish the data source to be connected according to the divided subject domain;

[0060] Configure the model for the database table in the selected data source and define the fields as atomic indicators and dimension dates;

[0061] Create new indicators and dimensions to be analyzed based on requirements, and define tags to associate with related fields;

[0062] The indicators, dimensions, and labels defined by the subject domain are added to the dataset so that they can be passed to the LLM model as a retrieval dataset when making queries.

[0063] The present invention is based on Figure 2 The query function architecture realizes conversational data query analysis based on LLM and semantic model. The system includes intelligent analysis module, semantic modeling module and knowledge base module.

[0064] Among them, the intelligent analysis module includes: a dialogue question and answer function, in which the user can enter the questions they want to query and analyze. The system calls the LLM module and the semantic analysis module to generate a query statement and execute it to return the data required by the user.

[0065] Among them, the semantic modeling module includes several functions: subject domain configuration, semantic model configuration, indicator management, dimension management, label management, and data set management. The subject domain configuration mainly divides the subject domain according to different subject applications, which can narrow the data query scope and perform data queries more accurately; the semantic model configuration can select the database table for model configuration, and the fields can be defined as atomic indicators and dimension dates; indicator management can add new indicators, which can be composite indicators and can be defined as formulas or connections; label management can define fields as labels, and searches can be performed based on labels; data set management can add indicators, dimensions, labels, etc. defined in the subject domain to the data set, and all data set definitions can be passed to LLM as prompts when querying.

[0066] Among them, the knowledge base module includes: LLM, which is a pre-trained large prediction model that can accurately understand the meaning of the user's input question and generate the semantic query SQL required by the user based on the prompt words and the knowledge base content.

[0067] like Figure 3 As shown, the specific steps of using the method of the present invention to perform conversational data query are as follows:

[0068] First, you need to pre-configure the semantic model, steps 1-4, so that the system can understand the data more accurately and return the query results accurately; after defining the semantic model, users can perform conversational data query analysis, steps 5-11.

[0069] 1. Define subject domain: Establish the data source to be connected and divide the subject domain according to different subject applications to narrow the data query scope and perform data query more accurately;

[0070] 2. Semantic model configuration: Select the database table in the data source connection to configure the model. The fields can be defined as atomic indicators and dimension dates.

[0071] 3. Define indicators, dimensions, and labels: You can create new indicators and dimensions to be analyzed as needed, and define labels to associate with related fields;

[0072] 4. Dataset management: Indicators, dimensions, tags, etc. defined in the subject domain can be added to the dataset. When querying, all dataset definitions can be passed to LLM as prompts;

[0073] 5. Users conduct data query and analysis: Users enter the questions they want to query and analyze in the dialog box, or they can directly ask and answer questions recommended by historical questions;

[0074] 6. Retrieve similar questions in the knowledge base: Retrieve similar question-answer pairs in the vector database using methods such as cosine similarity and Euclidean distance;

[0075] 7. Send data sets and similar questions as prompts to LLM: Send relevant data sets and similar questions as prompts to LLM based on keywords;

[0076] 8. LLM generates semantic SQL: LLM uses the large model capability to generate corresponding semantic SQL based on user questions and prompt words;

[0077] 9. Parse the semantic SQL into query SQL according to the semantic model: The semantic SQL cannot be executed directly. The field name in the SQL needs to be the information of the indicator defined in the semantic model. Parse the semantic SQL into executable query SQL according to the pre-defined;

[0078] 10. Execute SQL and return the query results to the user: Execute SQL in the analysis database and return the query results to the user;

[0079] 11. Display analysis content based on query result icons: Designed tables and relationships can be exported as physical models, SQL scripts can be exported or directly generated into the database by establishing a data source connection. The database supports a variety of mainstream databases and automatically adapts to different database types and dialects.

[0080] Preferably, the method further comprises:

[0081] The pre-set questions and SQL pair knowledge obtained after vectorization using the Text2vec method are stored in the vector database, and the vector database is enriched according to user queries.

[0082] Preferably, the method further comprises:

[0083] When the data to be analyzed will affect the business database and the analysis efficiency in the business database is not high, the data to be analyzed is extracted to the analysis database for analysis;

[0084] The query results are automatically displayed in a chart, and a multi-dimensional drill-down query is performed based on the query results, and the multi-dimensional drill-down query results are displayed.

[0085] Combination Figure 2 As shown, in the present invention, the knowledge base module also includes: a vector database and an analysis database. The vector database stores pre-set questions and SQL pair knowledge, which is stored after vectorization using methods such as Text2vec, and can enrich the knowledge base according to user queries; the analysis database stores the data to be analyzed by the user, and the data can be stored in the original business database. If the query analysis may affect the business database or the analysis efficiency of the original business database is not high, the data can be extracted to the analysis database for query analysis.

[0086] The intelligent analysis module also includes: the chart display module automatically displays the returned data in charts, showing the analysis results more intuitively, and the format of the displayed chart can be customized; multi-dimensional drilling can drill down and display the returned results according to the dimensions, and the intelligent recommendation function can recommend relevant questions and answers based on the user's questions, and can automatically add the answers judged by the user to the knowledge base, so that the question and answer system becomes more accurate with use.

[0087] The key points of the present invention are: 1. Integrate LLM and semantic model, expose consistent data semantics by building a unified semantic data model, incorporate data semantics (such as business terms, column values, etc.) into prompt words, so that LLM can better understand semantics to reduce hallucinations; unload the generation of advanced SQL syntax (such as connections, formulas, etc.) from LLM to the semantic layer to reduce complexity. 2. Vectorize the storage of built-in questions and query SQL pairs in the knowledge base, calculate the similarity of questions through cosine similarity, Euclidean distance and other methods when querying, and more accurately find similar questions, thereby improving the accuracy of SQL generation. 3. Add the correct answers to user queries to the knowledge base through conversational queries to improve the accuracy of answers.

[0088] The conversational data query and analysis method based on LLM and semantic model provided by the present invention has the following advantages: (1) The results obtained by performing data query and analysis in a conversational manner are more accurate, thereby improving the availability of ChatBI; (2) The indicator management function can not only enhance the accuracy of SQL generation, but also manage indicators in a unified manner and provide data services to the outside through API; (3) The combination of semantic model and LLM has lower cost. Since the complexity of large model SQL generation is reduced, only LLM with small parameters is needed to realize the generation of data query SQL, which greatly reduces the system application cost.

[0089] Figure 4 FIG. 4 is a schematic diagram of a data query analysis system 400 based on LLM and semantic model according to an embodiment of the present invention. Figure 4 As shown, the data query analysis system 400 based on LLM and semantic model provided by the embodiment of the present invention includes:

[0090] The target query statement parsing unit 401 is used to parse the target query statement input by the user to obtain a retrieval data set corresponding to the target query statement;

[0091] A similar question determining unit 402 is used to search similar questions in a vector database based on the target query statement to obtain similar questions;

[0092] A semantic SQL statement acquisition unit 403 is used to analyze the search data set and similar questions using a large language model LLM to acquire a semantic SQL statement;

[0093] The query result acquisition unit 404 is used to parse the semantic SQL statement into a query SQL statement by using a semantic model, and return a query result to the user based on the SQL query statement.

[0094] Preferably, the formal problem determination unit further comprises:

[0095] Based on the target query statement, similar questions are searched in a vector database through a cosine similarity or Euclidean distance system to obtain similar questions.

[0096] Preferably, the system further comprises: a configuration unit, configured to:

[0097] Divide the subject domain according to the subject application, and establish the data source to be connected according to the divided subject domain;

[0098] Configure the model for the database table in the selected data source and define the fields as atomic indicators and dimension dates;

[0099] Create new indicators and dimensions to be analyzed based on requirements, and define tags to associate with related fields;

[0100] The indicators, dimensions, and labels defined by the subject domain are added to the dataset so that they can be passed to the LLM model as a retrieval dataset when making queries.

[0101] Preferably, the system further comprises:

[0102] The vector database determination unit is used to store the pre-set question and SQL pair knowledge obtained after vectorization using the Text2vec method in the vector database, and enrich the vector database according to the user query.

[0103] Preferably, the system further comprises:

[0104] An analysis database determination unit, used for extracting the data to be analyzed into the analysis database for analysis when the data to be analyzed will affect the business database and the analysis efficiency in the business database is not high;

[0105] A display unit, used to automatically display the query results in a chart; and to display multi-dimensional drill-down query results;

[0106] The multi-dimensional drill-down query unit is used to perform a multi-dimensional drill-down query based on the query result, and obtain the multi-dimensional drill-down query result.

[0107] The data query analysis system 400 based on LLM and semantic model in the embodiment of the present invention corresponds to the data query analysis method 100 based on LLM and semantic model in another embodiment of the present invention, which will not be described in detail here.

[0108] The invention has been described above with reference to a few embodiments. However, it is readily apparent to a person skilled in the art that other embodiments than the ones disclosed above are equally within the scope of the invention, as defined by the appended patent claims.

[0109] Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the technical field, unless explicitly defined otherwise therein. All references to "a / said / the [means, components, etc.]" are to be openly interpreted as at least one instance of said means, components, etc., unless explicitly stated otherwise. The steps of any method disclosed herein do not necessarily have to be performed in the exact order disclosed, unless explicitly stated otherwise.

[0110] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0111] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0112] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0113] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0114] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A data query and analysis method based on LLM and semantic model, characterized in that: The method comprises: Parsing the target query statement input by the user to obtain a retrieval data set corresponding to the target query statement; Retrieving similar questions in a vector database based on the target query statement to obtain similar questions; Analyze the retrieval data set and similar questions using a large language model (LLM) to obtain semantic SQL statements; The semantic SQL statement is parsed into a query SQL statement using a semantic model, and a query result is returned to the user based on the SQL query statement.

2. The method according to claim 1, characterized in that The method further comprises: Based on the target query statement, similar questions are searched in a vector database by using a cosine similarity or Euclidean distance method to obtain similar questions.

3. The method according to claim 1, characterized in that The method further comprises: Divide the subject domain according to the subject application, and establish the data source to be connected according to the divided subject domain; Configure the model for the database table in the selected data source and define the fields as atomic indicators and dimension dates; Create new indicators and dimensions to be analyzed based on requirements, and define tags to associate with related fields; The indicators, dimensions, and labels defined by the subject domain are added to the dataset so that they can be passed to the LLM model as a retrieval dataset when making queries.

4. The method according to claim 1, characterized in that: The method further comprises: The pre-set questions and SQL pair knowledge obtained after vectorization using the Text2vec method are stored in the vector database, and the vector database is enriched according to user queries.

5. The method according to claim 1, characterized in that The method further comprises: When the data to be analyzed will affect the business database and the analysis efficiency in the business database is not high, the data to be analyzed is extracted to the analysis database for analysis; The query results are automatically displayed in a chart, and a multi-dimensional drill-down query is performed based on the query results, and the multi-dimensional drill-down query results are displayed.

6. A data query and analysis system based on LLM and semantic model, characterized in that: The system comprises: A target query statement parsing unit, used to parse the target query statement input by the user to obtain a retrieval data set corresponding to the target query statement; A similar question determining unit, configured to retrieve similar questions in a vector database based on the target query statement to obtain similar questions; A semantic SQL statement acquisition unit, used for analyzing the retrieval data set and similar questions using a large language model LLM to acquire a semantic SQL statement; The query result acquisition unit is used to parse the semantic SQL statement into a query SQL statement by using a semantic model, and return the query result to the user based on the SQL query statement.

7. The system according to claim 6, characterized in that The formal problem determination unit further includes: Based on the target query statement, similar questions are searched in a vector database through a cosine similarity or Euclidean distance system to obtain similar questions.

8. The system according to claim 6, characterized in that The system further comprises: a configuration unit, configured to: Divide the subject domain according to the subject application, and establish the data source to be connected according to the divided subject domain; Configure the model for the database table in the selected data source and define the fields as atomic indicators and dimension dates; Create new indicators and dimensions to be analyzed based on requirements, and define tags to associate with related fields; The indicators, dimensions, and labels defined by the subject domain are added to the dataset so that they can be passed to the LLM model as a retrieval dataset when making queries.

9. The system according to claim 6, characterized in that The system further comprises: The vector database determination unit is used to store the pre-set question and SQL pair knowledge obtained after vectorization using the Text2vec method in the vector database, and enrich the vector database according to the user query.

10. The system according to claim 6, characterized in that The system further comprises: An analysis database determination unit, used for extracting the data to be analyzed into the analysis database for analysis when the data to be analyzed will affect the business database and the analysis efficiency in the business database is not high; A display unit, used to automatically display the query results in a chart; and to display multi-dimensional drill-down query results; The multi-dimensional drill-down query unit is used to perform a multi-dimensional drill-down query based on the query result, and obtain the multi-dimensional drill-down query result.

Images