We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.CL

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computation and Language

Title: Is Automated Topic Model Evaluation Broken?: The Incoherence of Coherence

Abstract: Topic model evaluation, like evaluation of other unsupervised methods, can be contentious. However, the field has coalesced around automated estimates of topic coherence, which rely on the frequency of word co-occurrences in a reference corpus. Contemporary neural topic models surpass classical ones according to these metrics. At the same time, topic model evaluation suffers from a validation gap: automated coherence, developed for classical models, has not been validated using human experimentation for neural models. In addition, a meta-analysis of topic modeling literature reveals a substantial standardization gap in automated topic modeling benchmarks. To address the validation gap, we compare automated coherence with the two most widely accepted human judgment tasks: topic rating and word intrusion. To address the standardization gap, we systematically evaluate a dominant classical model and two state-of-the-art neural models on two commonly used datasets. Automated evaluations declare a winning model when corresponding human evaluations do not, calling into question the validity of fully automatic evaluations independent of human judgments.
Comments: Accepted to NeurIPS 2021 (spotlight presentation). CR version
Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Cite as: arXiv:2107.02173 [cs.CL]
  (or arXiv:2107.02173v3 [cs.CL] for this version)

Submission history

From: Alexander Hoyle [view email]
[v1] Mon, 5 Jul 2021 17:58:52 GMT (2136kb,D)
[v2] Mon, 4 Oct 2021 23:50:01 GMT (2524kb,D)
[v3] Wed, 27 Oct 2021 21:37:06 GMT (2144kb,D)

Link back to: arXiv, form interface, contact.