Improving WordNet using Word Embeddings

The main objective of this paper is to create a proof of concept regarding the improvement of the human-generated database WordNet using computer-generated information from Word2Vec. Thus, we change the WordNet content, using the information from existing corpora. The main method used to achieve this goal is by comparing the results of path algorithms for computing the semantic similarities between WordNet concepts (such as Path, Wu and Palmer, or Leacock and Chodorow similarities), with cosine similarity between the Word2Vec vectors of the same concepts. One way to improve WordNet is by adding new concepts from the Word2Vec corpus which have strong connections with existing words from WordNet. Another way to improve it is by updating its existing connections to underline semantic change. Our experimental results prove that the method we propose may be used to improve the number of concepts and the quality of links between synsets in WordNet, creating a more meaningful semantic resource.