A Novel Hybrid Methodology of Measuring Sentence Similarity

The problem of measuring sentence similarity is an essential issue in the natural language processing area. It necessary to measure between sentences accurately. Sentence task finding semantic symmetry two sentences, regardless word order and context words. There are many approaches similarity. Deep learning methodology shows a state-of-the-art performance fields used lot measurement methods. However, field, considering structure or that makes up also important. In this study, we propose combined with both deep method lexical relationships. Our evaluation metric Pearson correlation coefficient Spearman coefficient. As result, proposed outperforms current on KorSTS standard benchmark Korean dataset. Moreover, it performs maximum 65% increase than only using methodology. Experiments show our generally results better those model.