Development of a Reading Material Recommendation System Based on a Multi-expert Knowledge Acquisition Approach

Abstract

In English courses, it is very important to assign proper articles to individual students for training their reading ability. This study proposes an innovative approach for developing reading material recommendation systems by eliciting domain knowledge from multiple experts. An experiment has been conducted to evaluate the performance of the approach; moreover, a comparison on the existing approaches is given to show the advantages of applying the innovative approach.

Knowledge-Based Recommendation Systems

“…The recommendation is produced by computing the similarity between the user's preferences and services descriptions. Hsu et al (2009Hsu et al ( , 2010 propose a reading material recommendation system. The domain knowledge of various experts is elicited in order to propose articles to individual students so that they can train their reading ability.…”

“…As we can see, the range of KBRS varies depending mainly on the nature of the Knowledge Base, which influences the recommendation strategy. While some authors propose a knowledge base allowing a (conversational) case-based recommendation technique (Aktas et al, 2004;Burke, 1999;Chattopadhyay et al, 2012;Göker & Thompson, 2000;Khan & Hoffmann, 2003;Lee, 2004bLee, , 2004aLee & Lee, 2007;Yuan et al, 2013); other use a knowledge base containing formalized expert knowledge (Hsu et al, 2010(Hsu et al, , 2009Rosenfeld et al, 2013), a domain ontology (Ajmani et al, 2013;Blanco-Fernandez et al, 2008Carrer-Neto et al, 2012;Garcìa -Crespo et al, 2009;Kaminskas et al, 2012;), or a database (Ghani & Fano, 2002;Martinez et al, 2008;Towle & Quinn, 2000) and apply a similarity measure to generate a recommendation. Despite the differences between the recommendation methodologies, we can identify two common elements to each of the KBRS considered here: a Knowledge Base and a User Profile.…”

“…Some authors propose a preliminary step. This step is usually about preparing the data or ensure that all the necessary requirements for the completion of the methodology are satisfied (Hsu et al, 2009(Hsu et al, , 2010.…”

“…Other authors give indication on the means and ends of the knowledge acquisition step. Some methodologies make use of knowledge acquisition approaches that are already known to be effective (Hsu et al, 2009(Hsu et al, , 2010. While others develop and use their own approach (Blanco-Fernandez et al, 2008Garcìa-Crespo et al, 2009;Lee, 2004aLee, , 2004bRosenfeld et al, 2013).…”

“…Finally, when the knowledge base consists of formalized expert knowledge or a domain ontology, then the recommendation generation depends on the nature of the knowledge. If the knowledge can be subject to quantitative analysis, then a similarity measure is usually applied to provide the user with a recommendation (Blanco-Fernandez et al, 2011;Garcìa-Crespo et al, 2009;Hsu et al, 2010Hsu et al, , 2009Lee, 2004aLee, , 2004bMartinez et al, 2008). Otherwise, a match is made between the user problem and the stored knowledge to produce advice (Blanco-Fernandez et al, 2008).…”