Experimental results of semantic similarity methods
Description
This dataset collects data used for the comparison of the following seven semantic similarity methods: 1. SemSimp with a manual taxonomy and human-annotated documents. 2. SemSimAI with an automatic taxonomy and automatically annotated documents using BERTopic. 3. SemSimAI with an automatic taxonomy and automatically annotated documents using recursive k-means clustering with Word2Vec embeddings. 4. SemSimAI with an automatic taxonomy and automatically annotated documents using recursive k-means clustering with TF-IDF vectors. 5. Cosine similarity with document representations from TF-IDF vectors. 6. Cosine similarity with document representations from BERT embeddings. 7. Cosine similarity with document representations from Word2Vec embeddings.
Files
Institutions
- Istituto di analisi dei sistemi ed informatica Antonio Ruberti Consiglio Nazionale delle Ricerche
- ENEA Agenzia Nazionale per Le Nuove Tecnologie l'Energia e lo Sviluppo Economico Sostenibile