An Active Learning Based Support Tool for Extracting Hints of Tourism Development from Blog Articles

TL;DR: A tool to help analysts make tourism development ideas while reading blog articles using a support vector machine (SVM) and an active learning method that provides better results than the simple learning method.

Abstract: The present paper proposes a tool to help analysts make tourism development ideas while reading blog articles. Since reading the entire text of an article is time consuming, it is useful to extract from the blog articles significant sentences that are relevant to tourism development. The proposed tool extracts such sentences using a support vector machine (SVM) and an active learning method. In the first learning step, the proposed tool is trained using corpora that include hint-tags. The analyst then provides target blog articles to the tool and receives sentences as the results of the SVM classification. Some of these sentences are analyzed manually in order to annotate new hint-tags. In the second learning step, both the original corpora and the annotation results are used. Finally, the analyst reads plausible sentences extracted from the second classification of the target articles. In the experiments, we confirmed that the proposed active learning method provides better results than the simple learning method.