An image recommender system that suggests appropriate images in creation of self-learning items for abstract nouns

This article acquaints its readers with a research paradigm that has been designed to extract appropriate (synonymously educational) images of abstract nouns. The ongoing investigation principally purposed to assist L2 learners/educators by recommending appropriate images in the creation of vocabulary learning items for abstract nouns. Hence, a prototype version of an image recommender system that purposed to assist the users in the recommendation of appropriate images for 3-types of abstract nouns has been implemented and tested. The proposed system, in the process of learning items creation, allows the users to select their own preferred image if the system recommended topmost image is not satisfactory. Our study presumed that still images having physical or concrete existence can be addressed as appropriate learning resources in the representation of abstract nouns that firstly, represent social contexts between human, secondly, related to feeling and emotion, and thirdly, state social or religious belief. Authors agree that due to huge cultural influences and multiple behaviors of abstract nouns, this hypothesis may be a matter of debate. Therefore, to assess the images, an image evaluation experiment has been conducted with 20 participants who are actively engaged in foreign language acquisition. A post-hoc analysis of Tukey’s test revealed the significant difference (P=0.04) of our systemrecommended images over Yahoo-suggested images in learners’ considerations as appropriate image resources to memorize new vocabulary.