JASPER: Journal Article Selection Program for Non-native English Readers

Typically, reading a journal article can be timeconsuming, mainly for non-native English readers, because academic writing usually uses complicated vocabulary and sentences. Therefore, this paper proposes a Journal Article Selection Program for Non-native English Readers (JASPER) for selecting journal articles from abstracts using scanning and skimming techniques. JASPER employs linear searching as a scanning technique and a novel multi-layer Latent Dirichlet Allocation (LDA) as a skimming technique. It automatically classifies journal articles into multi-layer topics and selects only articles with related topics to reduce the number of articles readers must read. JASPER is evaluated in terms of accuracy and efficiency using journal articles on Computer Science topics. It achieved an average of 82.62% of the F-measure. It can also reduce the number of journal articles by an average of 98.68%. Therefore, JASPER can practically reduce the number of journal articles for non-native English readers.