Course Concept Expansion in MOOCs with External Knowledge and Interactive Game

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 4292–4302 Florence, Italy, July 28 - August 2, 2019. c⃝2019 Association for Computational Linguistics 4292 Course Concept Expansion in MOOCs with External Knowledge and Interactive Game Jifan Yu1,2, Chenyu Wang3, Gan Luo1,2 ,Lei Hou1,2∗, Juanzi Li1,2, Zhiyuan Liu1,2, Jie Tang1,2 1Dept. of Computer SCi.& Tech., Tsinghua University, China 100084 2KIRC, Institute for Artificial Intelligence, Tsinghua University, China 100084 3Shenyuan Honors College, Beihang University, China 100083 {yujf18@mails.,luog18@mails.}tsinghua.edu.cn {houlei@,lijuanzi@,liuzy@,jietang@}tsinghua.edu.cn wangchenyu@buaa.edu.cn Abstract As Massive Open Online Courses (MOOCs) become increasingly popular, it is promising to automatically provide extracurricular knowl- edge for MOOC users. Suffering from se- mantic drifts and lack of knowledge guidance, existing methods can not effectively expand course concepts in complex MOOC environ- ments. In this paper, we first build a novel boundary during searching for new concepts via external knowledge base and then uti- lize heterogeneous features to verify the high- quality results. In addition, to involve human efforts in our model, we design an interac- tive optimization mechanism based on a game. Our experiments on the four datasets from Coursera1 and XuetangX2 show that the pro- posed method achieves significant improve- ments(+0.19 by MAP) over existing methods. The source code3 and datasets4 have been pub- lished. 1 Introduction Self-determination theory was first formally pro- posed by Deci and Ryan in (Deci et al., 1991), suggesting that educators should support students in autonomously discovering and learning course- related knowledge. In fact, in addition to the con- cepts taught in course, many related concepts are also worthy of learning. Figure 1 shows a real example from Coursera in Data Structure course. When the concept Binary Search Tree is taught, some other concepts, including its similar struc- tures (Heap), applications (Sorting and Priority Queue) and advanced researches (Tango Tree5) ∗corresponding author 1https://www.coursera.org/ 2http://www.xuetangx.com/ 3Source Codes: https://github.com/thukg/ concept-expansion-kg 4Datasets: http://moocdata.cn 5It is an online binary search tree that achieves an O(log log n) competitive ratio. (Demaine et al., 2007) also benefit students for further course understand- ing. However, these concepts are not available without specifical mention, especially in the era of Massive Open Online Courses (MOOCs). In Figure 1: An example of “out-of-teaching” concepts in the course “Data Structure and Algorithm”. MOOCs, teachers need to keep a moderate length of the course to face with thousands of students with various backgrounds (Jordan, 2015), mak- ing it infeasible to manually pick out these helpful concepts. Therefore, there is a clear need to auto- matically identify course-related concepts, so that the students can easily acquire additional knowl- edge and achieve better educational outcomes. Although much work concerned with extract- ing course concepts from teaching materials (Kay and Holden, 2002) or course subtitles (Pan et al., 2017b) has been done, the research in finding the concepts absent in course materials, which we call Course Concept Expansion, has not been ex- plored. Despite abundant work on related top- ics, including concept expansion or set expan- sion (Wang and Cohen, 2007; Wang et al., 2015; …

Content truncated to 500 words