Abstractive Text Summarizer Using Attention Machanism

ive Text Summarizer Using Attention Machanism 1st Ashwini N. More Department of Computer Engineering Shrama Sadhana Bombay Trust’s College Of Engineering and Technology Jalgaon, India. [email protected] 3rd Yash R. Mahajan Department of Computer Engineering Shrama Sadhana Bombay Trust’s College Of Engineering and Technology Jalgaon, India. [email protected] 2nd Sakshi P. Mahajan Department of Computer Engineering Shrama Sadhana Bombay Trust’s College Of Engineering and Technology Jalgaon, India. [email protected] 4th Yashodhan Patil Department of Computer Engineering Shrama Sadhana Bombay Trust’s College Of Engineering and Technology Jalgaon, India. [email protected] Abstract— Abstractive text summarization is the task of generating a headline or a short summary consisting of a few sentences that captures the salient ideas of an article or a passage. We use the adjective ‘abstractive’ to denote a summary that is not a mere selection of a few existing passages or sentences extracted from the source. In this work, we will create the model of abstractive text summarization using Attentional Encoder Decoder Recurrent Neural Networks. Neural sequence-to-sequence models have provided a viable innovative approach for abstractive text summarization (meaning they are not restricted to simply selecting and rearranging passages from the original text). However, these models have two shortcomings: they are liable to reproduce factual details inaccurately, and they tend to repeat themselves. In this work we will used standard Long Short-Term Memory (LSTM) sequence-to-sequence attentional model. This method utilizes a local attention model for generating each word of the summary conditioned on the input sentence. While the model is structurally simple, it can easily be trained end-to-end and scales to a large amount of training data. We will apply our model to the Amazon-fine-food-review dataset. We will evaluate the reconstructed paragraph using standard metrics like ROUGE, showing that neural models can encode texts in a way that preserve syntactic, semantic, and discourse coherence. We propose several features that address problems which are coming in existing system in summarization that data visualization , Audio input, etc. and emitting words that are rare or unseen at training time.