Educational exploration along the shortest paths in conceptual networks based on co-occurrence, language ability levels and frequency ranking

Abstract

We propose a new computational method to support learning that relies on adaptiveexploration of the shortest paths in conceptual networks that have been formed based on cooccurrencesof concepts in suitable text samples and selecting concepts corresponding todesired language ability levels and frequency ranking. Relying on Google Web 1T 5-gramdatabase we have built a conceptual co-occurrence network reaching the coverage of 3018unique concepts and 54 610 unique pairs of co-occurring concepts thus approximatelymatching with a vocabulary size suggested to be enough for sufficient comprehension andwith the highest language ability levels of English Vocabulary Profile. Our method offers tothe learner recommendations about suitable exploration paths along the shortest connectingpaths between the concepts belonging to a desired learning topic vocabulary, computed withYen's algorithm. By indicating for each concept the language ability level and the frequencyranking position in British National Corpus enables to prioritize such shortest paths ofconcepts that most best match the current suitable comprehension level of the learner. Ourpreliminary experiment showed that the method can pedagogically intuitively supportcumulative adoption of knowledge in the context of study entities belonging to a corecurriculum. Relying on our research we are opening a free educational resource athttp://www.freelearningpath.org that enables learners and educators to get adaptive guidancefor exploring a desired educational content.