WildVis: Open Source Visualizer for Million-Scale Chat Logs in the Wild
Yuntian Deng, Wenting Zhao, Jack Hessel, Xiang Ren, Claire Cardie, Yejin Choi
2024-09-06
Summary
This paper talks about WildVis, an open-source tool designed to help researchers analyze large-scale chat logs quickly and effectively.
What's the problem?
As more real-world conversation data becomes available, it’s challenging for researchers to manually go through millions of chat logs to study how users interact with chatbots. The sheer volume of data makes it impractical to examine each conversation individually.
What's the solution?
WildVis addresses this issue by providing an interactive platform that allows users to search and visualize large datasets of chat logs based on various criteria. The tool uses optimizations like search index construction and data compression to ensure that users can interact with the data quickly, even when dealing with millions of entries. The authors also demonstrate WildVis's usefulness through case studies that explore different aspects of chatbot interactions.
Why it matters?
This research is important because it enables researchers to gain insights from vast amounts of conversation data without being overwhelmed by the volume. By making it easier to analyze user-chatbot interactions, WildVis can help improve chatbot design and functionality, ultimately enhancing user experiences in various applications.
Abstract
The increasing availability of real-world conversation data offers exciting opportunities for researchers to study user-chatbot interactions. However, the sheer volume of this data makes manually examining individual conversations impractical. To overcome this challenge, we introduce WildVis, an interactive tool that enables fast, versatile, and large-scale conversation analysis. WildVis provides search and visualization capabilities in the text and embedding spaces based on a list of criteria. To manage million-scale datasets, we implemented optimizations including search index construction, embedding precomputation and compression, and caching to ensure responsive user interactions within seconds. We demonstrate WildVis's utility through three case studies: facilitating chatbot misuse research, visualizing and comparing topic distributions across datasets, and characterizing user-specific conversation patterns. WildVis is open-source and designed to be extendable, supporting additional datasets and customized search and visualization functionalities.