Lexical similarity based query-focused summarization using artificial immune systems

Abstract

Query Focused Summarization has been explored mostly with statistical or graph based methods which haven’t utilised semantic similarity between words. Graph Based methods which use sentence to sentence comparisons do not utilize lexical relations between words fully due to entailing complexity of finding relationship among all words. Lexical Chaining Methods which are used in Generic Text Summarization systems also utilize only a limited set of word types such as nouns. They do not utilize the full potential of Semantic Similarity measures by overlooking sentence to sentence comparisons. We propose a novel method for Query Focused Summarization which makes full use of semantic relationships between sentences arising out of relationships between their constituent words by using Artificial Immune Systems to compare the sentences thereby reducing the complexity. Experiments show the potential of the approach to be used in situations with large input data.