Data Analysis Support System for Cross-Table Recommendation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data analysis systems struggle to recommend analysis target items across multiple tables without relying on table definition information, particularly in the railway industry, where data from different systems is difficult to integrate and analyze due to the lack of standardized data specification sheets.
Innovation Solution
A data analysis support apparatus and system that classifies data items into types, generates a relationship network between systems and data tables, and calculates scores for analysis target candidates, allowing users to select analysis targets without needing table definition information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data items are combined variously to find analysis target groups, then analysis capability is improved, but the importance and burden of this work increase as the number of data items increases
Solution Approach 1:
The system automatically extracts data items and generates analysis target groups without requiring manual intervention. The data specification sheet is automatically parsed to extract table definitions, column information, and data relationships, eliminating the need for analysts to manually combine data items
Solution Approach 2:
Manual data analysis work is replaced by an automated system that uses machine learning models to calculate relevance scores between data items and generate analysis target groups based on the extracted data specifications
2Loss of information
If data specification sheets are used to understand data, then data understanding is improved, but it becomes impossible or difficult when specification sheets are not ready or unavailable
Solution Approach 1:
A natural language processing intermediary is introduced between the raw data and the analysis system. The system extracts and structures information from data specification sheets when available, creating an intermediate representation that can be used even when the original specification sheets are unavailable or difficult to access
Solution Approach 2:
The system performs preliminary extraction and structuring of data specifications in advance, storing the extracted information in a usable format. This preliminary action allows the system to work with pre-extracted data characteristics even when the original specification documents are not immediately accessible
3Ease of operation
If analysis target items are recommended based on relevance calculation, then analysis target selection is improved, but analysis target items across multiple tables cannot be recommended
Solution Approach 1:
The relevance calculation mechanism is enhanced to work universally across single-table and multi-table scenarios. The system extracts data relationships from specification sheets and uses these relationships to calculate relevance scores between data items even when they belong to different tables, enabling cross-table analysis target recommendations
4Adaptability or versatility
If analysis is performed across multiple tables, then analysis scope is improved, but table definition information must be extracted from DB which increases complexity
Solution Approach 1:
Table definition information is extracted and stored in advance from data specification sheets before analysis is performed. This preliminary extraction creates a ready-to-use repository of table schemas, column definitions, and data relationships that can be directly utilized for multi-table analysis without requiring complex runtime extraction from databases
Data Source
AI summary
A data analysis support apparatus includes a relationship network generation section that analyzes a relationship between operating systems, a relationship between operation data tables, a relationship between data items possessed by the operation data tables and a relationship between data values possessed by records of the operation data tables and stores them, as a relationship network; a data item classification section that classifies data items that become a data analysis target into a first data type based on an actual value and a second data type based on a planned value; an analysis data table generation section that generates and accumulates an analysis data table to be used for data analysis; a data model generation section that generates, as a data model, a data item group that allows data analysis in combination; and an analysis target item presentation section that recommends a data item to be made an analysis target.