Extracti̇ve text summari̇zation by gray wolf optimization algorithm and classification of abstracts with deep learning

Today, especially in the internet environment, textual data is increasing rapidly. It is getting more and more difficult to reach the desired piece in these texts, which have become a large data set. Automated text summarization techniques play a major role in extracting relevant information from these big data. In this thesis, the intuitive Gray Wolf Optimization (GWO) algorithm is proposed as inferential text summarization technique. Clustering ability GWO algorithm has been tested with single text summarization application. In the summarization system, statistical keyword extraction methods such as sentence order, word length and cluster extraction of GWO are combined. The summary system was also tested with the K-means clustering algorithm and the results were measured with the ROUGE evaluation metric. In accordance with the results obtained in the study tested with BBC News data set consisting of 2225 news articles, GWO algorithm was observed to perform quite well. The summarized texts are classified with one of the deep learning methods, Long- Short term memory (LSTM) networks. Experimental results showed that the classification success of the LSTM network increased in the summaries created with the GWO algorithm. GWO algorithm is presented as a new approach for inferential summarization methods.