Design of a Location-based Word Recommendation System Based on Association Rule Mining Analysis
Mohammad Nehal Hasnine, Gökhan Akçapınar, Brendan Flanagan, Kousuke Mouri, Hiroaki Ogata, Keiichi Kaneko, Noriko Uosaki · 2019
Association rule mining is the process of discovering new knowledge and non-trivial patterns hidden in textual datasets. This approach is often used by learning analytics researchers and practitioners for exploring unique types of data collected from various educational environments. By leveraging association rule mining, this study aims at providing a solution to a critical challenge that most of the location-based informal learning tools face namely, the limited scopes to learn new words. Hence, the purpose of this work is to design a word recommendation system that can analyze a learner's previously learned words in order to suggest new vocabulary that can be acquired in a particular learning location. We designed a study and explored a dataset for uncovering the insight of language learners' learning behaviors. 20 EFL (English as Foreign Language) learners' data, whose learning location was Tokushima region of Japan, is analyzed by using association rule mining technique. The result indicates that-First, in informal learning context, EFL learners' word learning interests vary much from each other although the learning location is same; and Second, a word recommendation system can be introduced by using association rule mining analysis; however, the possibility of getting highly associated words are low compared with random words.