Query Oriented Extractive-Abstractive Summarization System (QEASS)

Abstract

This work proposes a query oriented extractive-abstractive summarization system where the query is synthesized and expanded from the novel details provided by the patent analyst and the domain ontology. Since the search and patent document retrieval using the formulated semantic query alone will not satisfy the user requirements, this work filters and summarizes the retrieved document set both extractively and abstractively. Summarization makes use of deep learning techniques as their structure mimics the human brain. The proposed work was evaluated using Google patent dataset. The retrieval results of semantic query expansion using domain ontology are compared with Google Prior-art search query results and WordNet based semantic query expansion retrieval sets. The summarization results of the retrieved document sets are compared with the human summaries.