Hybrid approach for Word Sense Disambiguation in Marathi Language

Word Sense Disambiguation (WSD) is the capacity to compute the appropriate meaning of a word in a particular context. Among the top languages, one of the languages is Marathi, which is also the mother tongue of state Maharashtra, India. Marathi language has many words, phrases which have the multiple uses and can be used at various places in the sentences interpreting different meaning. This ambiguity is well identified by humans manually, but for machine or artificial intelligence, it is not that easy and work is required to develop the models which can identify the accurate meaning interpretations for such words used in a sentences. Therefore, it is inspiring to create a corpus of Marathi that would communicate the accurate meaning of a word with ambiguity. Some of the existing WSD systems use various supervised learning algorithms which classifies the meaning of the word in the given sentence. The approaches that are available are limited to the small size datasets and also have the accuracy issues. Here, the efforts are taken to improve the accuracy of such WSD system for the Marathi language using combination of knowledge based approach and machine learning technique such as Support Vector Machine (SVM). The model is trained with the real time dataset prepared by taking the help of Marathi language experts and real time environment is created for better results and working of the system.