SSE: Semantic Sentence Embedding for learning user interactions

Semantics is an integral component in NLP. Semantics provides a meaningful view about the meaning of the language. The meaning of the text is susceptible to the negative words present in them. Thus the impact of semantics in NLP in turn increases the impact of negative words in the sentence. It plays crucial role in shaping the semantics of any sentence. Although all the sub-areas of natural language processing try to find out the impact of negative words on their implementation, the effect of these words are not considered in the word embedding process. Being the foremost step in NLP the omission of these words in embedding can affect the representation of the language. The aim is to propose a model which produces word embeddings by considering the negative words. Although most NLP implementing systems consider all negative words this paper focuses on the word ‘not’ which is followed by an adjective. The antonym for the adjective is identified and is then replaced in the corpus. The model is achieved by replacing the corpus where negative words and this corpus is fed to a three layer neural network to form the embeddings. The proposed technique proves an accuracy of 93.6% in retrieving the relevant answer to the user.

[1]  Ali Farhadi,et al.  Bidirectional Attention Flow for Machine Comprehension , 2016, ICLR.

[2]  Iván Sánchez Rodríguez.  Text similarity by using GloVe word vector representations , 2017 .

[3]  Jeffrey Dean,et al.  Efficient Estimation of Word Representations in Vector Space , 2013, ICLR.

[4]  Wael Hassan Gomaa,et al.  A Survey of Text Similarity Approaches , 2013 .

[5]  Zuhair Bandar,et al.  Sentence similarity based on semantic nets and corpus statistics , 2006, IEEE Transactions on Knowledge and Data Engineering.

[6]  Hui Wang,et al.  R-Trans: RNN Transformer Network for Chinese Machine Reading Comprehension , 2019, IEEE Access.

[7]  Curt Burgess,et al.  Explorations in context space: Words, sentences, discourse , 1998 .

[8]  Naoaki Okazaki,et al.  Sentence Extraction by Spreading Activation through Sentence Similarity , 2003 .

[9]  Feiyue Ye,et al.  Sentences similarity analysis based on word embedding and syntax analysis , 2017, 2017 IEEE 17th International Conference on Communication Technology (ICCT).

[10]  Zhoujun Li,et al.  Response selection from unstructured documents for human-computer conversation systems , 2018, Knowl. Based Syst..

[11]  Nan Yang,et al.  Context-Aware Answer Sentence Selection With Hierarchical Gated Recurrent Neural Networks , 2018, IEEE/ACM Transactions on Audio, Speech, and Language Processing.

[12]  Jeffrey Pennington,et al.  GloVe: Global Vectors for Word Representation , 2014, EMNLP.

[13]  Jun Zhao,et al.  How to Generate a Good Word Embedding , 2015, IEEE Intelligent Systems.

[14]  M. Dolores del Castillo,et al.  Human-inspired semantic similarity between sentences , 2015, BICA 2015.

[15]  Xiaolong Wang,et al.  Content-Oriented User Modeling for Personalized Response Ranking in Chatbots , 2018, IEEE/ACM Transactions on Audio, Speech, and Language Processing.

[16]  James L. McClelland,et al.  Mechanisms of Sentence Processing: Assigning Roles to Constituents of Sentences , 1986 .

[17]  Vijay K. Mago,et al.  Calculating the similarity between words and sentences using a lexical database and corpus statistics , 2018, ArXiv.

[18]  Rong Jin,et al.  Understanding bag-of-words model: a statistical framework , 2010, Int. J. Mach. Learn. Cybern..

[19]  Zhoujun Li,et al.  Response selection with topic clues for retrieval-based chatbots , 2016, Neurocomputing.

[20]  Jeffrey Dean,et al.  Distributed Representations of Words and Phrases and their Compositionality , 2013, NIPS.