Paper Title
Text Summarization Using Multi-Relational Graphs and Text Generation Techniques
Abstract
Topic of text summarization has been a hot topic in the field of natural language processing and will continue to
play a hot research based topic role in our information overloaded world where people don’t have time to go through the
whole text available to them whether the information is from the internet or the daily newspaper,most of people want a brief
summary of the text without having to read through the whole document and our solution to the chosen problem statement
may provide a solution to the information overload problem. Text summarization in natural language processing focuses on
the process of using algorithms to briefly describe the large bodies of text without compromising on the gist of the document
with the summary being accurate,unbiased and emphasizing on the key facts of the text. Our technique proposes a Graph
convolution network algorithm and clustering based summarization approach. The proposed approach consists of three main
steps: Pre-processing of the dataset.Then Building Graph Convolution network (GCN) model to learn the syntactic
representation for a document. The selective attention mechanism is used to extract salient information in semantic and
structural aspects and generate an extractive summary. Same model is applied on the dataset to obtain an extractive
summary. The novelty of our proposed solution to the problem is to show that the summarization result not only depends on
the sentence features, but also depends on the sentence similarity measure. GCN takes into account not only the semantic
and structural aspects of the text document and provides the summary. The evaluation metric and the result obtained by
training the GCN model on the BBC dataset shows that our proposed approach can improve the performance compared to
other traditional methods of summarization.
Keywords - GCN, BBC