Explainable NLQ-based Visual Interactive System: Challenges and Objectives

Publisher:
ACM
Publication Type:
Conference Proceeding
Citation:
ICCA '22: Proceedings of the 2nd International Conference on Computing Advancements, 2022, pp. 420-425
Issue Date:
2022-03-10
Filename Description Size
3542954.3543014.pdfPublished version606.29 kB
Adobe PDF
Full metadata record
Nowadays, visual interactive systems (Vis) are attracting more attention in research and industries because of their effectiveness in conveying information. Additionally, to make rational decisions based on extracted data, Vis is critical for identifying and comprehending trends, outliers, and patterns in data. Existing research has employed a broad range of methodologies to yield visualization insights into certain decision-making systems, allowing participants to perceive a specific problem from a wide range of viewpoints. However, there are still enough scopes to design a new Vis where some systematic techniques are required to visualize the data with proper explanations. In this regard, we analyze several existing works and observe a surge of research interest in the new realm of explainable and NLQ-based Vis. In this paper, our main goal is to present a novel idea for designing an explainable NLQ-based Vis named-ExNLQVis. Therefore, (i) we aim to discuss a proposed NLQ-based Vis that will follow a deep learning-based NLP approach to extract necessary information from user inputs, make visual-respective decisions, and generate appropriate visualizations based on the preceding decisions. (ii) we extend our prior model to an explainable visualization model that not only accurately visualizes data but also explains why it appears depending on the natural language query (NLQ). To accomplish this system, we consider several challenges and objectives and briefly discuss our proposed method accordingly. We also provide the implementation and evaluation guidelines to establish our system.
Please use this identifier to cite or link to this item: