# Leveraging word embeddings to enhance co-occurrence networks: A statistical analysis

## TL;DR

This paper studies how adding semantic connections to word networks affects their usefulness for analyzing short texts.

## Contribution

The study provides insights into how virtual edges from word embeddings impact network metrics for different text properties.

## Key findings

- Adding virtual edges improves some network metrics for short texts but harms others like clustering coefficient.
- Stopwords influence the statistical properties of networks enriched with semantic edges.
- The results help choose suitable network metrics based on text length and task requirements.

## Abstract

Recent studies have explored the addition of virtual edges to word co-occurrence networks using word embeddings to enhance graph representations, particularly for short texts. While these enriched networks have demonstrated some success, the impact of incorporating semantic edges into traditional co-occurrence networks remains uncertain. In this study, we investigate two key statistical properties of text-based network models. First, we assess whether network metrics can effectively distinguish between meaningless and meaningful texts. Second, we analyze whether these metrics are more sensitive to syntactic or semantic aspects of the text. Our results show that incorporating virtual edges can have both positive and negative effects, depending on the specific network metric. For instance, the informativeness of the average shortest path and closeness centrality improves in short texts, while the clustering coefficient’s informativeness decreases as more virtual edges are added. Additionally, we found that including stopwords affects the statistical properties of enriched networks. Our results, derived from enriching networks with FastText embeddings, offer a guideline for identifying the most appropriate network metrics for specific applications, based on typical text length and the nature of the task.