The paper proposes a modified version of Differential Evolution (DE) algorithm and optimization criterion function for extractive textsummarization applications. Cosine Similarity measure has been used to cluster sim...
详细信息
ISBN:
(纸本)9781479980840
The paper proposes a modified version of Differential Evolution (DE) algorithm and optimization criterion function for extractive textsummarization applications. Cosine Similarity measure has been used to cluster similar sentences based on a proposed criterion function designed for the text summarization problem, and important sentences from each cluster are selected to generate a summary of the document. The modified Differential Evolution model ensures integer state values and hence expedites the optimization as compared to conventional DE approach. Experiments showed a 95.5% improvement in time in the Discrete DE approach over the conventional DE approach, while the precision and recall of extracted summaries remained comparable in all cases.
暂无评论