Natural Language Processing

K.Kaku, S.Anda, M.Morita

Efficient supporting of various activities using NLP.

writing assistance for thesis, assistance of review browsing, lyric analysis

data mining, deep leaning, title generation, summarization, information extraction, author classification

In the task of writing papers, there are a number of difficult task for students who are not used to writing papers. We research two tasks that assist these students. The first task is to assist those who create paper titles. In this task, we research on a system that generates titles from abstracts by applying summarization methods. The second task is to assist those who decide which conferences and journals to submit papers to. In this task, we research on assistance in choosing conferences and journals to which the paper should be submitted from information on the paper, such as title and abstract.
Product reviews on e-commerce websites are used by users for the purpose of efficiently gathering information of interest. However, product reviews are often not broken down by the parts or aspects mentioned, which makes information gathering difficult. We research the assistance of review browsing by classifying and extracting reviews by the parts mentioned and their aspects.Lyrics are considered to be influenced by components such as singers, lyricists and themes. Lyric analysis in this research is aimed at extracting the features contained in lyrics. In this task, we analyse lyrics by evaluating the results of classifying the dataset considering the relationship between lyricist and singer with a lyricist classifier based on BERT.