Profile Summarization From Semantic Data Using Deep Neural Network Model
6 Pages Posted: 15 Aug 2019 Last revised: 3 Sep 2019
Date Written: August 15, 2019
Abstract
By querying a Database, the results are often triples. It is a time consuming job to understand the triples and identify the relationship between the m and also the information contained in it. So think of an automated system that could convert these triples into English language. Wont it be a boon to this hustling world?? That is just the thing proposed in this paper. Summarization plays a crucial role in this hectic world. Everyone is busy and hence no time to spend to read long paragraphs and understands their meaning. Automatic text summarization is a challenging task that involves lot of steps. One of the hurdles is to understand the meaning of the text and then process it. So in this work, semantic data in the form of RDF Triples is being given and using deep learning approaches, a concise and coherent summary is being generated. This work primarily focuses in the domain of profile summarization. A study about systems that perform RDF to natural language conversion is been done and inferred information that models to develop the proposed system. The limitation of such models is that only intra triple relationship is conserved. And this limitation is solved in the proposed model by using a knowledge graph constructed from RDF triples.
Keywords: Sequence-to-sequence model, Natural Language Generation, RNN, Knowledge Graph, RDF Triples
Suggested Citation: Suggested Citation