Abstract—With an excess of information in recent times, sound information retrieval is the need of the hour. Document Classification, where a document is classified as being under one of a number of predefined categories, is the foundation of an efficient and effective Information Retrieval system. Once information has been retrieved, the next step is unearthing the relevant and essential information. Query-based Multi- Document Summarization will do just that. In this paper, we analyze the different variants of the k Nearest Neighbors (kNN) Classification Algorithm and from them design the CAST Algorithm for Classification, which, as precision and recall results will show, performs better in most cases. For document summarization, we analyze and improvise on a Hyper graph based algorithm. Further, we design and describe the CAST Algorithm for Summarization and show that it performs well for Query-based Multi-Document Summarization.
Index Terms—Document Classification, Multi-Document Summarization, Query-based summary.
Suzanne D’Silva is with the Veermata Jijabai Technological Institute, Matunga, Mumbai – 400019, India (email: suzanne.d1411@gmail.com).
Neha Joshi is with the Veermata Jijabai Technological Institute, Matunga, Mumbai – 400019, India (email: nehaj_26@yahoo.co.in).
Sudha Rao is with the Veermata Jijabai Technological Institute, Matunga, Mumbai – 400019, India (email: raosudha89@gmail.com).
Sangeetha Venkatraman is with the Veermata Jijabai Technological Institute, Matunga, Mumbai, India (email: sangeetha2089@gmail.com).
Seema Shrawne is with the Veermata Jijabai Technological Institute, Matunga, Mumbai – 400019, India (email: scshravane@vjti.org.in).
Cite: Suzanne D’Silva, Neha Joshi, Sudha Rao, Sangeetha Venkatraman, and Seema Shrawne, "Improved Algorithms for Document Classification &Query-based Multi-Document Summarization," International Journal of Engineering and Technology vol. 3, no. 4, pp. 404-409 , 2011.
Copyright © 2008-2024. International Journal of Engineering and Technology. All rights reserved.
E-mail: ijet_Editor@126.com