Importance of Dataspace Embeddings when Evaluating Text Clustering Methods

Alain Lelu (1) , Martine Cadot (2)

1 UFC - Université de Franche-Comté
2 MULTISPEECH - Speech Modeling for Facilitating Oral-Based Communication

Fair evaluation of text clustering methods needs to clarify the relations between 1)pre-processing, resulting in raw term occurrence vectors, 2)data transformation, and 3)method in the strict sense. We have tried to empirically compare a dozen well-known methods and variants in a protocol crossing three contrasted open-access corpora in a few tens transformed dataspaces. We compared the resulting clusterings to their supposed "ground-truth" classes by means of four usual indices. The results show both a confirmation of well-established implicit combinations, and good performances of unexpected ones, mostly in spectral or kernel dataspaces. The rich material resulting from these some 600 runs includes a wealth of intriguing facts, which needs further research on the specificities of text corpora in relation to methods and dataspaces.

Mots clés: evaluation method; method comparison; text clustering; K-Means; Normalized Matrix Factorization; Latent Dirichlet Allocation; hierarchical clustering; linkage method; spectral clustering; graph partition; kernel clustering; Salton tfidf; Okapi tf-idf; chi-square metrics; Laplacian spectral decomposition; correspondence analysis; kernel expansion; evaluation method; evaluation method

Domaines: Autre [cs.OH]; Traitement du texte et du document; Autres [stat.ML]

Alain Lelu, Martine Cadot. Importance of Dataspace Embeddings when Evaluating Text Clustering Methods. Data Analysis and Rationality in a Complex World, In press. ⟨hal-03053176v2⟩