The article ‘Labeled Interactive Topic Models’ by Kyle Seelman, Mozhi Zhang, and Jordan Boyd-Graber discusses the enhancement of neural topic models through user interactivity. The authors correct the lack of interactivity in neural topic models like the Embedded Topic Model (ETM) by allowing users to label a topic. The labeled topics are then updated to align the topic words closely with the label. This process enables users to refine topics based on their information needs. The authors also developed an interactive interface for users to relabel the topic models. They conducted a human study to evaluate the method and found that user labeling improves document rank scores, helping to find more relevant documents.
Publication date: 17 Nov 2023
Project Page: Not provided
Paper: https://arxiv.org/pdf/2311.09438
