Topic Modeling on Podcast Short-Text Metadata
This work addresses the challenge of organizing and navigating podcast collections more effectively for users and platforms, though it is incremental as it builds on existing topic modeling techniques.
The researchers tackled the problem of discovering relevant topics from short podcast metadata by proposing a new strategy to leverage named entities in a Non-negative Matrix Factorization framework, which led to improved topic coherence over baselines on datasets from Spotify, iTunes, Deezer, and a new catalog.
Podcasts have emerged as a massively consumed online content, notably due to wider accessibility of production means and scaled distribution through large streaming platforms. Categorization systems and information access technologies typically use topics as the primary way to organize or navigate podcast collections. However, annotating podcasts with topics is still quite problematic because the assigned editorial genres are broad, heterogeneous or misleading, or because of data challenges (e.g. short metadata text, noisy transcripts). Here, we assess the feasibility to discover relevant topics from podcast metadata, titles and descriptions, using topic modeling techniques for short text. We also propose a new strategy to leverage named entities (NEs), often present in podcast metadata, in a Non-negative Matrix Factorization (NMF) topic modeling framework. Our experiments on two existing datasets from Spotify and iTunes and Deezer, a new dataset from an online service providing a catalog of podcasts, show that our proposed document representation, NEiCE, leads to improved topic coherence over the baselines. We release the code for experimental reproducibility of the results.