Content-based multi-document summarizer

In the contemporary scenario of information overload, the most challenging task before the computing society is to devise methods for efficient representation and timely delivery of relevant information in short and precise chunks to the end user. Since the vast majority of web documents is written in some natural language, tools and techniques of Natural Language Processing (NLP) assume significance here. In this paper, we present a text processing system, which consists of modules for phrasal indexing, content-based classification, and domain-specific summarization. The primitive operation of string matching has been optimized and implemented by a specialized co-processor, which interfaces with other modules of the system and improves the overall speed. The system has been evaluated with real world documents as its inputs.