/Word-Sense-Embeddings

Sense Embeddings for Word and Relational Similarity.

Primary LanguageJupyter Notebook

Word Sense Embeddings

The task is to implement Sense Embeddings for Word and Relational Similarity. I implemented a CBOW model using word2vec.

Results

Spearman's correlation = 0.52

Compatibility

Python3.X,  Gensim

Sense Embeddings

The sense embedddings can be found in https://drive.google.com/open?id=1vi_AcGDJhKWtCdbgXme4DXyMKTj-2my0

Notes

In this project, two datasets were merged for training: EuroSense from http://lcl.uniroma1.it/eurosense/ And Semantically Enriched Wikipedia from http://lcl.uniroma1.it/sew/ .   

This dataset was used for testing: WordSimilarity-353 from http://www.cs.technion.ac.il/~gabr/resources/data/wordsim353/ .