Improving Lesk by Incorporating Priority for Word Sense Disambiguation

There exists a scarcity of work done in choosing the optimal WSD algorithm as applicable for different situations and even less work has been done in applying WSD techniques to low-resourced languages like Bengali. This paper presents three modified versions of the Lesk algorithm along with the original version, as applied to Bengali. The results obtained from experimentation, particularly our dependency tree based Lesk variant proves to be highly robust for handling the sense disambiguation task in Bengali.