2016 International Conference on Big Data and Smart Computing (BigComp) (2016)
Hong Kong, China
Jan. 18, 2016 to Jan. 20, 2016
Chanida Deerosejanadej , Department of Computer Engineering, King Mongkut's University of Technology Thonburi, Bangkok, Thailand
Phond Phunchongharn , Theoretical and Computational Science Center and Department of Mathematics, King Mongkut's University of Technology Thonburi, Bangkok, Thailand
Tiranee Achalakul , Department of Computer Engineering, King Mongkut's University of Technology Thonburi, Bangkok, Thailand
This paper presents an opinion summarization approach based on sentence clustering and sentence selection. To automatically generate a comprehensive and non-redundant summary, review sentences are grouped together with a modified genetic algorithm (GA) before selecting a representative from each group. The modified-GA has three different components compared to the original, namely fitness function, gene reassignment operation and encoding technique. Apart from sentence clusters, our modified genetic algorithm also provides probabilistic membership degrees of each sentence for each cluster to indicate how similar the sentence is to other members of the cluster. Later, these degrees can be taken into account to generate a comprehensive summary in the sentence selection process. Since the core of this work resides in the sentence clustering process, our modified genetic algorithm is evaluated by comparing with other conventional methods. The results reveal that our algorithm significantly outperforms the others in both accuracy and execution time. Therefore, our approach should produce more comprehensive and less redundant summary.
Genetic algorithms, Clustering algorithms, Encoding, Sociology, Statistics, Genetics, Semantics
C. Deerosejanadej, P. Phunchongharn and T. Achalakul, "A sentence clustering framework for opinion summarization using a modified genetic algorithm," 2016 International Conference on Big Data and Smart Computing (BigComp)(BIGCOMP), Hong Kong, China, 2016, pp. 269-272.