A Personal Conversation Assistant Based on Seq2seq with Word2vec Cognitive Map

DOI 機関リポジトリ

この論文をさがす

抄録

type:Article

WeChat is one of social network applications that connects people widely. Huge data is generated when users conduct conversations, which can be used to enhance their lives. This paper will describe how this data is collected, how to develop a personalized chatbot using personal conversation records. Our system will have a cognitive map based on the word2vec model, which is used to learn and store the relationship of each word that appears in the chatting records. Each word will be mapped to a continuous high dimensional vector space. Then the sequence-to-sequence framework (seq2seq) will be adopted to learn the chatting styles from all pairs of chatting sentences. Meanwhile, the traditional one-hot embedding layer will be replaced with our word2vec embedding layer in the seq2seq model. Furthermore, an autoencoder of seq2seq architecture is trained to learn the vector representation of each sentence, then the cosine similarity between model generated response and the pre-existing response in test set can be evaluated , and the distance with principal component analysis (PCA) projection can be also displayed. As a result, our word2vec embedded seq2seq model significantly outperforms the one-hot embedded one.

収録刊行物

キーワード

詳細情報 詳細情報について

問題の指摘

ページトップへ