Design and implementation of a recommender system for online vocabulary learning
No Thumbnail Available
URL
Journal Title
Journal ISSN
Volume Title
Perustieteiden korkeakoulu |
Master's thesis
Ask about the availability of the thesis by sending email to the Aalto University Learning Centre oppimiskeskus@aalto.fi
Authors
Date
2016-10-27
Department
Major/Subject
Human-Computer Interaction and Design
Mcode
SCI3020
Degree programme
Master's Programme in ICT Innovation
Language
en
Pages
72
Series
Abstract
Recommender systems are used to select, within a wide catalog, a limited number of products (such as films, books, or general items) which are presented to the user based on their preferences. This master's thesis work has been carried out in the company Kielikone. The company owns an online learning service called Sanakirja.fi, which helps in studying and teaching foreign languages through the use of wordlists. Wordlists are collections of terms, and the users can play with them in order to memorize words. The goal of this thesis is the analysis and development of recommendation algorithms adapted to infer the users' taste and identify categories for each wordlists in order to achieve targeted recommendation. There exist various types of recommendation systems, which differ mainly in the way how the recommendations for the users are produced. In this master's thesis work we focus on the integration of three techniques, namely, collaborative filtering, content-based recommendations and the combination of both, in a hybrid approach. We show that by creating a text classifier we are able to categorize wordlists based on the topic of the words contained in them. Based on the literature, we implemented also a collaborative filtering algorithm, which is able to obtain accurate recommendation based on the users' implicit ratings. In the end, we combined these two approaches to obtain wordlist recommendations, which are related to the topics the user has expressed his interest in, and having at the same time a good number of diversified and serendipitous results.Description
Supervisor
Kaski, SamuelThesis advisor
Laaksonen, JormaStén, Liisa
Keywords
recommendation system, collaborative filtering, matrix factorization, content-based recommendation, word learning