Due to the development of the Internet, the continuous increase in online learning resources has made it possible for students to self-study. This project looks forward to using artificial intelligence deep learning methods to generate quizs to assist students' self-study and teacher teaching. In this three years project, three modules will be constructed that can be used by students and teachers in the process of self-study, and improve the efficiency of learning and teaching. The main goal is to build an automatic proposition system in the first year, a personalized mutual aid learning match in the second year, and an automatic learning profolio generation with blockchain storage in the third year. The first year is mainly in the process of learning. One research argued that quiz can improve the learning performance. However, the quiz often requires the teacher to design. The process of human propositions is quite time-consuming and labor-intensive, and it is not easy for students to obtain simulated test questions related to the course. Usually, they can only rely on the reference in the reference book. Practice and self-assessment with practice questions, publicly-published question banks or past archaeological questions. The goal of the first year plan is to use multi-source textbooks to find the teacher's teaching focus and select sentences through a topic model that can find the underlying semantics of the document. The problem construction phase is different from the traditional deep learning seq2seq architecture. This research will use Feature-rich Encoder adds extra features to the word, and proposes the concept of Topic-embedding Decoder to add a holistic document theme, hoping to generate smoother questions that match the content of the chapter. In the second year, the project will be shifted to the problem coaching staff. Most students will still encounter bottlenecks in self-study. They need problem solvers to assist them. Traditionally, problem solvers are teachers. In the second year project, the goal is to use collaborative learning, using the collaborative filter (CF) method, to find out students who are close in learning direction, and match with each other to become a recommended target, but the biggest problem with similar methods is that they may not willing to assist students with problems, this situation will be solved using the third year plan. In the third year, project aims to automatically generate report of the learning profolio and keep in blockchain by analyzing the history of being a mentor and learner during the collaborative learning process. Use the well-known record to improve the problem of insufficient motivation for the second year after the match. And the use of text generation technology to automatically generate learning profolio to reduce the lack of human records of learning process data.
|Effective start/end date||20-08-01 → 21-07-31|
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.