Abstract
In this paper, a method of text summarization based on deep learning is proposed. The method combines the topic information into the Bidirectional LSTM sequence to sequence model and provide thematic and context alignment information into the deep learning architecture to better deal with the long-term dependencies of text summarization, so as to avoid missing the topic words and the relationship between words in the syntactic structure of source text, help the model generate more coherent, informative, and more relevant summary of the topic information. This method is called a Bidirectional LSTM Sequence-to-Sequence Model based on Topic Fusion (TF-BiLSTMS2S). In the experiments, we use the Large-scale Chinese Short Text Summarization Dataset (LCSTS) to evaluate the model, and the ROUGE index was used to evaluate the results. The experimental results show that the proposed model is effective and feasible for abstractive text summarization.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have