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

TL;DR

This paper introduces a combined visual and textual analysis tool that leverages automated highlights, LLM-generated insights, and visual cues to streamline anomaly detection in complex datasets, especially for financial fraud detection.

Contribution

It presents an integrated system that combines automated visual cues, textual summaries, and visual analytics to enhance exploratory data analysis and anomaly detection.

Findings

The tool effectively supports data exploration and anomaly detection.

Domain experts found the system helpful in identifying suspicious information.

The approach reduces information overload during analysis.

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.