Visibility Graph Representations of Word Embeddings for Text Classification
Organizations: Department of Computer Engineering, Faculty of Computer and Information Sciences, Karabük University, Karabük, Türkiye
Abstract
Text classifiers depend on representations that expose the distinctions between classes in a form the learning model can use, and a restricted vocabulary changes which representation is most effective. This paper studies natural visibility graphs as an intermediate representation for news topic classification under controlled vocabulary budgets. Word embeddings are averaged into a fixed-length document vector, the vector is read as an ordered numerical series, and its natural visibility graph is passed to a classifier either as an ordered node-degree vector or as an adjacency matrix. On the four-class AG News benchmark, chi-square selection retains 100 to 1,500 words, and the graph inputs are compared with direct embedding-based representations from continuous bag-of-words, skip-gram, GloVe, FastText and BERT. With a fully connected classifier, the degree representation is the most accurate configuration from 700 selected words onward; it reaches its highest accuracy of 91.24% at 1,000 words and exceeds the BERT-based baseline by 0.29 percentage points at 1,500 words. In a separate convolutional comparison, the adjacency input reaches 86.16% at 1,000 words, against 83.12% for the strongest static embedding. The results indicate that visibility-derived features can be competitive once the selected vocabulary provides adequate coverage. We discuss the roles of graph compression, coordinate ordering and classifier choice, and the controls needed to establish the reliability of the small differences observed.
Figures & tables
| Approach | Graph or representation | Learning setting |
|---|---|---|
| TextGCN [ 13 ] | Corpus word–document relations | Graph convolution |
| Onan [ 14 ] | Hierarchical linguistic graph | Contextual fusion and graph learning |
| Donabauer and Kruschwitz [ 15 ] | Contextual token graph | Graph-based text classification |
| Ghadiri et al. [ 20 ] | Voice visibility graph plus text | Multimodal fusion |
| This study | Visibility graph of a numerical document series | Degree or adjacency inputs |
| Category | Training | Test |
|---|---|---|
| World | 30,000 | 1,900 |
| Sports | 30,000 | 1,900 |
| Business | 30,000 | 1,900 |
| Science and Technology | 30,000 | 1,900 |
| Total | 120,000 | 7,600 |
| GloVe | FastText | Skip-gram | CBOW | BERT | Degrees | |
|---|---|---|---|---|---|---|
| 100 | 56.88 | 74.82 | 66.88 | 67.50 | 76.85 | 68.86 |
| 200 | 70.92 | 80.46 | 76.87 | 77.21 | 80.82 | 74.50 |
| 300 | 76.32 | 83.64 | 79.70 | 80.24 | 85.24 | 80.22 |
| 400 | 78.25 | 85.62 | 82.82 | 82.70 | 85.84 | 84.22 |
| 500 | 78.76 | 86.54 | 83.23 | 83.39 | 87.96 | 87.27 |
| 600 | 78.85 | 86.73 | 84.41 | 84.44 | 89.48 | 89.31 |
| GloVe | Skip-gram | CBOW | Graph | |
|---|---|---|---|---|
| 100 | 57.12 | 60.43 | 64.43 | 65.80 |
| 200 | 71.86 | 71.10 | 69.57 | 74.52 |
| 300 | 76.75 | 75.14 | 73.92 | 76.96 |
| 400 | 78.72 | 77.38 | 76.70 | 80.19 |
| 500 | 81.56 | 78.40 | 76.27 | 80.69 |
| 600 | 81.89 | 80.47 | 78.90 | 82.25 |
Text Classification
Text classification assigns text to categories such as topic, sentiment, intent, stance, or risk, supporting search, moderation, and decisions.
See all 239 papers