A preliminary study on semi-automatic construction of sense tagged corpus with Wordnet senses using semantic vector

A sense tagged corpus is important resources to word sense disambiguation and word sense identification. However, the sense tagging process should be conducted manually that it is a time consuming and expensive process. Therefore, the semi-automatic construction of sense tagged corpora is studied to reduce the cost of word sense tagging. This construction should contain two important phases that are word sense recommendation, which is done automatically, and word sense correction, which is done manually. This paper aims at word sense recommendation phase and presents the method of automatically identifying word sense with Wordnet senses using semantic vector. To evaluate the method, a component implementing word sense recommendation phase with latent semantic index, and then with word vector estimation, is tested with Wordnet and SemCor version 3.0 corpus.