Many-objective Extractive Document Text Summarization using NSGA-II

Extractive document text summarization plays an important role in obtaining relevant information from a large article. It finds application in social media analysis, news, legal documents and email summerization. In this manuscript this task is formulated as a many-objective optimization problem by simultaneously minimizing four objective functions: reciprocal of coverage of summery, redundancy of information, reciprocal of proper noun count score and reciprocal of numerical value count score. Along with these four objectives a fixed length of the summery is taken as a constraint. Popular Multi-objective Nondominated Sorting Genetic Algorithm-II (NSGA-II) is employed to perform the optimization task. The simulation is carried out on two case studies CNN-Daily mail data set. Comprative analysis has been carried out using text summarization with K Means, Fuzzy logic and Fuzzy C-Means method. Performance evaluation using precision shows better results with proposed NSGA-II, whereas using F-measure the results are second best among the four algorithms.