Why Didn't You Listen to Me? Comparing User Control of Human-in-the-Loop Topic Models

TL;DR

This paper compares different human-in-the-loop topic modeling approaches, evaluating their control and quality trade-offs through simulation experiments, and introduces a control metric to assess user satisfaction.

Contribution

It provides a systematic comparison of HLTM methods, extending existing frameworks with new evaluation metrics and insights into control versus quality trade-offs.

Findings

Informed prior methods offer better user control.

Constraints produce higher quality topics.

Control metric effectively measures user satisfaction.

Abstract

To address the lack of comparative evaluation of Human-in-the-Loop Topic Modeling (HLTM) systems, we implement and evaluate three contrasting HLTM modeling approaches using simulation experiments. These approaches extend previously proposed frameworks, including constraints and informed prior-based methods. Users should have a sense of control in HLTM systems, so we propose a control metric to measure whether refinement operations' results match users' expectations. Informed prior-based methods provide better control than constraints, but constraints yield higher quality topics.