Vec2GC -- A Graph Based Clustering Method for Text Representations

TL;DR

Vec2GC is a new unsupervised clustering method that uses graph community detection on text representations to effectively cluster terms or documents, supporting hierarchical clustering.

Contribution

The paper introduces Vec2GC, a novel end-to-end clustering algorithm leveraging graph community detection on learned text representations, applicable to various text corpora.

Findings

Effective clustering of terms and documents demonstrated.

Supports hierarchical clustering.

Applicable to unlabeled text data.

Abstract

NLP pipelines with limited or no labeled data, rely on unsupervised methods for document processing. Unsupervised approaches typically depend on clustering of terms or documents. In this paper, we introduce a novel clustering algorithm, Vec2GC (Vector to Graph Communities), an end-to-end pipeline to cluster terms or documents for any given text corpus. Our method uses community detection on a weighted graph of the terms or documents, created using text representation learning. Vec2GC clustering algorithm is a density based approach, that supports hierarchical clustering as well.