Human-Annotated Topic Model Evaluation and Topic Resources for Comparative Analysis

This dataset contains the topics extracted from various topic modeling techniques (LDA, BERTopic, and TopClust) along with human evaluations assessing their coherence. The evaluation process involved expert annotators providing qualitative assessments of topic interpretability and coherence. The dataset is structured to facilitate further research on topic modeling evaluation, including correlations between human judgments and automated coherence metrics.

It includes:

• The extracted topics from different models.

• Human-annotated coherence scores.

• Aggregated comparisons between models.