A Novel Unsupervised Learning Method for Word Sense Disambiguation using Word Vector

Word sense disambiguation has many applications in different fields. However, existing word sense disambiguation algorithms are mostly based on context and semantic term coverage, and usually do not consider the distance influence of words and ambiguous words in context. To this end, in this paper, a novel unsupervised learning method based on word vector is proposed. The vector is used to represent the context and the meaning, and the semantic similarity and the distribution frequency of the semantics of the fusion context and the meaning of the semantics are considered. The method is tested on SemEval-2010 Dataset and the results show that the method outperforms the state-of-the-art algorithms.

کلیدواژه ها:

Word Sense Disambiguation ، Word Vector ، Natural Language Processing ، Machine Translation