Host: Japan Society for Fuzzy Theory and Intelligent Informatics (SOFT)
Name : 35th Fuzzy System Symposium
Number : 35
Location : [in Japanese]
Date : August 29, 2019 - August 31, 2019
The purpose of this research is to summarize news articles based on word significance. Based on the collected news articles, we calculate the significance of the words of each sentence in the news article and extract three important sentences. We also consider whether the sentences are similar to the headline article. For the extracted sentences, a framework is created based on the subject, predicate, object, and object related terms. Summary sentences are created using the framework including the significant words. As a result of subject experiment, the naturality of Japanese and the tolerance as s summary was improved. There is a problem that the evaluation was low when the sentence is unnatural as Japanese even if the content of the summary is correct.
