© 2020 JETIR April 2020, Volume 7, Issue 4 www.jetir.org (ISSN-2349-5162) JETIR2004442 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org 1780 Construction of Concept Graph for Education Domain Roshankumar Pradhan 1 , Kaushal Tajane 2 , Akash Yadav 3 , Prof. Sonali Mane 4 1,2,3 Final year student, Department of Information Technology, BVCOE, Navi Mumbai. 4 Professor, Department of Information Technology, BVCOE, Navi Mumbai. Abstract — Inspired by vast applications of data graph and the ever increasing demand in education domain, we would propose a system, to automatically construct Concept graph for education domain. Concept graph is an integrated information repository that interlinks heterogeneous data from different domains. By leveraging on heterogeneous data (e.g., pedagogical data and learning assessment data) from the education domain, this technique first extracts concepts of subjects or courses, and so identifies the tutorial relations between concepts. More specifically, it adopts the Wikipedia web page crawling on pedagogical data to extract instructional concepts, and employs comparison between two chapters of the subject input data to spot the relations with educational significance. We detail all the above efforts through an exemplary case of constructing a demonstrative concept graph for a selected subject, where the educational concepts and their prerequisite relations are derived from curriculum standards and concept based performance data of scholars. It would not only be helpful in the Education sector but anywhere where learning is a task. Keywords- Concept Graph, Prerequisite relation, Knowledge graph, Educational Domain, Text Mining. I. INTRODUCTION Concept graph serves as an integrated information repository that interlinks heterogeneous data from different domains. Google’s Knowledge Graph is such a prominent example that represents real world entities and relations throu gh a multirelational graph. Existing generic knowledge graphs have demonstrated their advantages in supporting a large number of applications, typically including semantic search (e.g., Google’s Knowledge Panel), personal assistant (e.g. Apple’s Siri) and deep question answering (e.g., IBM’s Watson and Wolfram Alpha). However, those generic knowledge graphs usually cannot well support many domain-specific applications, because they require deep domain information and knowledge. Education is one of such domai ns, and in this work, we mainly focus on how the concept graph for education can be automatically constructed. In education domain, Concept graphs are often used for subject teaching and learning in school, where they are also called concept maps. Moreover, popular massive open online course (MOOC) platform such as Khan Academy, also adopt them for concept visualization and learning resource recommendation. Such Concept graphs are usually constructed by experienced teachers or domain experts in a manual way. However, such a manual construction process is actually time consuming and not scalable to large number of concepts and relations. What’s more, the number of courses and subjects grows fast on MOOC platforms, so it is much more difficult, or even impossible, to manually construct concept graphs for each new course. Students need prior knowledge for thorough understanding of educational content. This need imparts an implicit order in learning educational concepts. Determining this order requires significant human time and eff ort. Furthermore, relying on expert knowledge to determine this order is subject to inconsistencies due to ‘expert blind spot’, which means expert’s cognition and learner’s cognition on the same concept often do not well align. In other words, even the domain experts or experienced teachers may easily misunderstand learners’ cognitive process. As a result, those manually created Concept graphs can be suboptimal or misleading for learners. II. RELATED WORK Google’s knowledge graph, a variety of generic knowledge graphs, such as Reverb, Google Vault, Freebase, and Microsoft’s Probase, have been constructed by industry and academia, mainly utilizing data collected from the Internet. In educational domain, few studies focus on systematic construction of concept graphs[10], but there are some recent works investigating different relation extractions between certain known educational entities: Devendra et al.[1] automated induce of prerequisite structures of multiple units in a course, they propose a generic algorithm to use educational material and student activity data from heterogeneous sources to create a Prerequisite Structure Graph; Wang et al.[2] extract concepts hierarchies from the textbooks, extracts important concepts in each book chapter using Wikipedia as a resource and from this construct a concept hierarchy for that book; Liang et al.[3] recovers prerequisite relations from course dependencies, Wikipedia data is exploited to find prerequisite relations among universally shared concepts using both the Wikipedia article contents and their linkage structures; Chen et al[4] constructed knowledge graph for academic and online courses, the main concepts are extracted using neural network and data mining technique is used to find the prerequisite relation and Yu Lu et al.[5] constructed a system that build graph for online platforms. The most relevant work to our research is carried out by Carnegie Mellon University: The researchers utilize observed relations among courses to create a directed concept graph [6], but the relations are assumed to be known in advance.