Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

"Show Me What's Wrong!": Combining Charts and Text to Guide Data Analysis

Beatriz Feliciano, Rita Costa, Jean Alves, Javier Liébana, Diogo Duarte, Pedro Bizarro

TL;DR

A tool combining automated information highlights, Large Language Model generated textual insights, and visual analytics, facilitating exploration at different levels of detail in multi-dimensional datasets is presented, easing the identification of suspicious information.

Abstract

Analyzing and finding anomalies in multi-dimensional datasets is a cumbersome but vital task across different domains. In the context of financial fraud detection, analysts must quickly identify suspicious activity among transactional data. This is an iterative process made of complex exploratory tasks such as recognizing patterns, grouping, and comparing. To mitigate the information overload inherent to these steps, we present a tool combining automated information highlights, Large Language Model generated textual insights, and visual analytics, facilitating exploration at different levels of detail. We perform a segmentation of the data per analysis area and visually represent each one, making use of automated visual cues to signal which require more attention. Upon user selection of an area, our system provides textual and graphical summaries. The text, acting as a link between the high-level and detailed views of the chosen segment, allows for a quick understanding of relevant details. A thorough exploration of the data comprising the selection can be done through graphical representations. The feedback gathered in a study performed with seven domain experts suggests our tool effectively supports and guides exploratory analysis, easing the identification of suspicious information.

"Show Me What's Wrong!": Combining Charts and Text to Guide Data Analysis

TL;DR

A tool combining automated information highlights, Large Language Model generated textual insights, and visual analytics, facilitating exploration at different levels of detail in multi-dimensional datasets is presented, easing the identification of suspicious information.

Abstract

Analyzing and finding anomalies in multi-dimensional datasets is a cumbersome but vital task across different domains. In the context of financial fraud detection, analysts must quickly identify suspicious activity among transactional data. This is an iterative process made of complex exploratory tasks such as recognizing patterns, grouping, and comparing. To mitigate the information overload inherent to these steps, we present a tool combining automated information highlights, Large Language Model generated textual insights, and visual analytics, facilitating exploration at different levels of detail. We perform a segmentation of the data per analysis area and visually represent each one, making use of automated visual cues to signal which require more attention. Upon user selection of an area, our system provides textual and graphical summaries. The text, acting as a link between the high-level and detailed views of the chosen segment, allows for a quick understanding of relevant details. A thorough exploration of the data comprising the selection can be done through graphical representations. The feedback gathered in a study performed with seven domain experts suggests our tool effectively supports and guides exploratory analysis, easing the identification of suspicious information.
Paper Structure (17 sections, 5 figures)

This paper contains 17 sections, 5 figures.

Table of Contents

  1. Introduction
  2. Related Work
  3. Graphical representation of attributes
  4. Text summary
  5. Multiple levels of detail
  6. Highlight relevant information
  7. Design goals
  8. Visualization design
  9. Knowledge Area Console (A)
  10. Knowledge Area Text Summary (B)
  11. Knowledge Area Graphical Representation (C)
  12. Backend System
  13. Automated Risk-Assessment and Attribution
  14. Summary Generation
  15. Graphics Engine
  16. ...and 2 more sections

Figures (5)

  • Figure 1: Different states that a KA can have in the Console region.
  • Figure 2: Examples of text summaries for different KAs: alerted person (on the left) and alerted person activity (on the right).
  • Figure 3: Example of a text summary of a KA not flagged as fraudulent (on the left) and of a KA flagged as fraudulent (on the right).
  • Figure 4: Examples of graphical representations of different KAs: alerted person (on top) and alerted person activity (on the bottom).
  • Figure 5: Overview of the architecture of the system.