A prototype for sentence classification using statistical methods

Classification of sentences is a challenging problem in the field of natural language processing. We present a prototype for classifying the pairs of sentences using their syntactic and semantic relations. The sentences are classified into three classes — entailment, contradiction and neutral. The grammatical relations of the sentences are extracted and these relations are represented as word embeddings using global vector, Glove. The word to word semantics is found using Wordnet semantic relations. The prototype is based on statistical measures which was implemented using Python 3.6.9 version.