Data analysis device and computer-implemented method thereof

Abstract

A computer-based method for data analysis is provided, comprising: a command input step, which receives natural language commands from a user through a user interface and parses the natural language commands using a language model to determine at least one data lookup request within the natural language commands; a data point lookup step, which receives the data lookup request from the command input step, looks up at least one data point corresponding to the data lookup request, and outputs lookup parameters corresponding to the data point; a data acquisition step, which acquires at least one piece of data corresponding to the data point based on the lookup parameters; and a data analysis step, which proposes analysis suggestions based on the data.

Claims

1. A computer-implemented method for data analysis, executed by a processing module, the method comprising: an instruction input step, receiving natural language instructions from a user through a user interface; parsing the natural language instructions through a language model and determining at least one data lookup request in the natural language instructions; a data point lookup step, receiving the data lookup requests from the instruction input step, looking up at least one data point corresponding to the data lookup requests, and outputting lookup parameters corresponding to the data points; a data acquisition step, acquiring at least one piece of data corresponding to the data points based on the lookup parameters; detecting whether the data acquired by the data acquisition step contains at least one error; an instruction correction step, in response to detecting that the data contains the errors, re-parsing the natural language instructions and correcting the errors, and outputting a corrected natural language instruction; in response to receiving the corrected natural language instruction from the instruction correction step, executing the data point lookup step and the data acquisition step; and a data analysis step, proposing analysis suggestions based on the data.

2. The computer implementation method for data analysis as described in claim 1, wherein the instruction correction step further comprises: providing instructional prompts corresponding to the instruction correction step to the language model to receive and parse the errors; providing a tool call function to access a data point lookup table; re-parseing the natural language instruction and correcting the errors through the language model, using thought chains and referring to the data point lookup table; and providing outline prompts corresponding to the instruction correction step to the language model to output the corrected natural language instruction according to a predefined format; wherein, The predefined format includes the natural language instruction and the modified natural language instruction.

3. The computer implementation method for data analysis as described in claim 1, wherein the data point lookup step further comprises: providing instructional prompts corresponding to the data point lookup step to the language model to obtain at least one semantic structure corresponding to the data lookup requirements; providing a tool call function to access a data point lookup table; using the language model, looking up the data points corresponding to the semantic structures according to the data point lookup table; and providing outline prompts corresponding to the data point lookup step to the language model to output the lookup parameters according to a predefined format; wherein, This data point lookup table stores predefined identification information corresponding to different data points.

4. The computer implementation method for data analysis as described in claim 3, wherein the data point lookup step further comprises: performing the data point lookup step a predefined number of times and recording multiple sets of data points corresponding to the predefined number of data point lookup steps; comparing whether the multiple sets of data points are the same as each other; and in response to the multiple sets of data points being different from each other, displaying the multiple sets of data points through the user interface and prompting options for selecting one of the multiple sets of data points.

5. The computer implementation method for data analysis as described in claim 1, wherein the data extraction step further comprises: providing an instructional prompt corresponding to the data extraction step to the language model to receive and parse the search parameters; providing a tool call function to perform the following steps: accessing a structured database and extracting data from the structured database based on the search parameters; generating at least one chart based on the data; and generating at least one summary based on the data and / or the chart through the language model.

6. The computer implementation method for data analysis as described in claim 5, wherein the data analysis step further includes: providing instructional prompts corresponding to the data analysis step to the language model to receive the summaries; providing the tool call function to access a knowledge base and obtain relevant information corresponding to the summaries; and, through the language model, proposing analysis suggestions based on the relevant information of the summaries.

7. The computer implementation method for data analysis as described in claim 1 further comprises: determining whether the natural language instruction contains at least one knowledge question; and, in response to the natural language instruction containing the knowledge questions, performing the data analysis step to provide analysis suggestions corresponding to the knowledge questions.

8. The computer implementation method for data analysis as described in claim 7, wherein the data analysis step further comprises: providing instructional prompts corresponding to the data analysis step to the language model to obtain at least one semantic structure corresponding to the knowledge questions; providing a tool call function to access a knowledge base and obtain relevant information corresponding to the semantic structures; and, through the language model, proposing analysis suggestions corresponding to the knowledge questions based on the relevant information of the semantic structures.

9. A data analysis device, comprising: a user interface configured to receive natural language commands from a user; a storage module configured to store a data point lookup table and a structured database; and a processing module coupled to the user interface and the storage module, and configured to perform: a command input step, parsing the natural language command through a language model and determining at least one data lookup request in the natural language command; a data point lookup step, receiving the data lookup requests, looking up at least one data point corresponding to the data lookup requests from the data point lookup table, and outputting lookup parameters corresponding to the data points; a data retrieval step, retrieving data corresponding to the data points from the structured database according to the lookup parameters; and detecting whether the data retrieved by the data retrieval step contains at least one error; and a command correction step, in response to detecting that the data contains the errors, re-parsing the natural language command and correcting the errors, and outputting a corrected natural language command. In response to receiving the corrected natural language instruction from the instruction correction step, the data point search step and the data retrieval step are executed; the data analysis step proposes analysis suggestions based on the data and displays the analysis suggestions through the user interface.

Images