Sentence Semantic Similarity Using Dependency Parsing
Tanmay Vakare, Kshitij Verma, Vedant Jain · 2019
We introduce a novel approach to calculate the semantic similarity between sentences. Unlike conventional approaches, we leverage dependency grammar representations of the sentences and semantic word similarity methods to find the semantic similarity between the sentences. While conventional methods can efficiently extract syntactic structure, they fail to determine the semantic relation between the sentences. Furthermore, we also explore a similar approach which was originally proposed for extractive multi-document summarisation problem. Finally, we evaluate the newly proposed approach, similar approach and the conventional approaches on the data set which is considered as the gold standard for semantic similarity problems.