DT2VIS: A Focus+Context Answer Generation System to Facilitate Visual Exploration of Tabular Data

Area of Science:

• Human-Computer Interaction
• Data Visualization
• Artificial Intelligence

Background:

• Visual analysis dialog systems offer natural language interfaces for data analysis.
• Existing systems often prioritize query understanding over answer generation and exploration.
• User queries may not fully capture all analysis intentions.

Purpose of the Study:

• To propose a focus+context answer generation approach for simultaneous insight and contextual information.
• To design and implement a query recommendation algorithm for in-depth data exploration.
• To introduce the DT2VIS system for enhanced visual data analysis.

Main Methods:

• Developed a focus+context answer generation strategy to handle incomplete user queries.
• Implemented a query recommendation algorithm leveraging user historical data.
• Integrated these components into the DT2VIS system.
• Conducted a comparative study to evaluate system performance.

Main Results:

• DT2VIS effectively generates answers with contextual information.
• The query recommendation algorithm facilitates deeper user exploration.
• Comparative analysis demonstrated DT2VIS's efficiency in achieving analysis goals.
• The system aids users in reaching their data analysis objectives.

Conclusions:

• The focus+context approach addresses limitations in current visual analysis dialog systems.
• DT2VIS enables more efficient and insightful data exploration through natural language.
• The system empowers users to achieve their analysis goals effectively.