Android based News Summarization

Project Title

Android based News Summarization

Project Area of Specialization Artificial IntelligenceProject Summary

Android-based Automatic Text summarization, is to shorten the amount of text in the document without losing the essence, is one of the most challenging and a time-consuming activity because of the complexity of natural language processing involved with different kind of data. There are a number of techniques available to perform automatic text summarization and are classified broadly into extractive and abstractive. We are going to implement a hybrid approach i.e. Extractive + Abstractive using a novel attention mechanism. In this method, firstly salient sentences from the document and rewrite them in order to get final summaries. Nowadays, people use the internet to find information through information retrieval (IR) tools such as Google, Yahoo, Bing and so on. However, with the exponential growth of information on the internet, information abstraction or a summary of the retrieved results has become necessary for users. In the current era of information overload, text summarization has become an important and timely tool for a user to quickly understand the large volume of information.

Project Objectives Summarization is an important challenge of natural language understanding. The main objective is to provide abstraction-based summarization near to perfect like produced by the human expert and to produce a condensed representation of an input text that captures the core meaning of the original. As humans are generally good at abstract summary generation type of task as it involves first understand the meaning of the source document and then distilling the meaning and capturing silent details in the new description. Project Implementation Method Extractive summarization usually shows a better performance comparing to the abstractive approach especially with respect to ROUGE. In this project, we are using an approach i.e. Extractor + Pointer-Generation Network. In this model, a unified framework that tries to leverage the sentence-level salient information from an extractive model and incorporate them into an abstractive model (a pointer-generative network). More formally, inspired by the hierarchical attention mechanism by replacing the attention distribution in the abstractive model with a scaled version. It is the same as a sentence-level salient score of the sentence at word position and decoding step. Benefits of the Project One of the advantages of this project is that it is using both extractive and abstractive approach as the extractive approaches is that which can summarize source articles by extracting salient snippet and sentences directly from these documents, while abstractive approaches rely on word-level attention mechanism to determine the most relevant words to the target word at each decoding step. Technical Details of Final Deliverable The model which will be used in this project is the Pointer-generator network which is the addition in the seq2seq framework with attention model. The dataset which will be used is CNN/Dailymail dataset which contains the news articles with highlights to train the model. The training strategies which are used are word-level and sentence-level training with Beam search summary generation algorithm. Final Deliverable of the Project Software SystemCore Industry OthersOther Industries IT Core Technology Artificial Intelligence(AI)Other Technologies OthersSustainable Development Goals Quality EducationRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	80000
Android Mobile	Equipment	1	70000	70000
For printing and stationary	Miscellaneous	10	1000	10000