Face-to-face classes had been replaced by online classes from primary schools to universities aro... more Face-to-face classes had been replaced by online classes from primary schools to universities around the world due to COVID-19. In a 4th year computer science course, we forwent conventionally separated lectures and programming laboratories lasting 2 hours each and switched to alternating mini-lectures and student exercises several times in a 2-hour timeslot. Students reported that they learned better with improved motivation in this new class format. Students also indicated that they were mostly neutral about whether the face of professor was shown in online lecture.
Some non-native speakers struggle with lectures conducted in English. It may be hard for them to ... more Some non-native speakers struggle with lectures conducted in English. It may be hard for them to pick out key messages when reading books or lecture notes. Microsoft PowerPoint has an “add notes” feature. We propose to use it to attach lecture notes to individual slides instead of compiling all lecture notes in a conventional book form. Students learn key points from slides and detailed explanation from slide-based lecture notes. A suitable learning approach, deep or shallow, can be chosen according to personal learning goal on a slide-by-slide basis. Student-centered learning is empowered to a high level of granularity in a blended learning environment. Lecturers may update slide-based lecture notes more effectively and efficiently than separate sets of slides and lecture notes.
Outcome-based teaching and learning emphasizes the explicit declaration of learning outcomes whic... more Outcome-based teaching and learning emphasizes the explicit declaration of learning outcomes which identify the tasks students are expected to be able to perform after completing the course, and to what standard. OBTL also requires the teaching, learning and assessment activities to align with the stated learning outcomes. We interviewed fifteen university instructors about their experience of teaching outcome-based computer science
Communications in computer and information science, 2020
As an important educational form, online learning has attracted millions of registered learners, ... more As an important educational form, online learning has attracted millions of registered learners, and a huge number of courses are available online. However, it is challenging for learners to identify appropriate courses from a large course pool due to the difficulties of mapping complex learning needs to the high-level course semantics. Several studies in the field of Natural Language Processing (NLP) have recently gained promising performance in capturing the semantic information. In this study, we use these NLP techniques to understand the semantics of learning needs and courses. Specifically, we model users’ historical course records as word sentences using skip-gram with negative sampling to obtain course semantics. Furthermore, we introduce Laplacian Eigenmaps as the objective function and integrate the course social tags and course-user interaction as penalty factors to fine-tune the course vectors, especially the courses of different categories but similar contexts. The result verifies that the proposed method is effective for recommending suitable courses for users.
During the past five years, the development of various technologies and the maturing of related e... more During the past five years, the development of various technologies and the maturing of related education industries have never ceased, and the Hong Kong government continually invests into the comprehensive establishment of technology-enhanced education systems, especially those for English education. Here we then meet the point where the efforts of the past five years call for a panoramic view and an in-depth analysis of people's attitudes toward technology-enhanced language learning. This study investigated a total of 56 in-service English teachers' acceptance of technology-enhanced language learning and teaching. These teachers were from 56 different primary schools in Hong Kong. The results presented an overview of the current situation of e-learning in Hong Kong English education, showing insights into the future TELL design and development of related programs.
Students in a year 4 computing course were given a choice of learning activities using different ... more Students in a year 4 computing course were given a choice of learning activities using different media: live lectures, on-demand recorded lectures, and PowerPoint slides with/without transcripts. Students preferred slides with transcripts more than the other learning media because it enables them to concentrate better. However, students perform equally well regardless of their preferred choice of learning medium. We found that students who studied regularly week after week performed better than students who studied erratically. Our twofold recommendations are for courses designed to provide multiple learning media for students to choose and to encourage students to study regularly rather than in an erratic pattern.
In the past decade, it has become popular to use instructional videos for teaching and learning i... more In the past decade, it has become popular to use instructional videos for teaching and learning in online and blended learning environments. While researchers have studied how the presence of an instructor in an instructional video affects learning effectiveness, the influence of an instructor's visual familiarity toward students' learning is unclear. This experimental study explored how face familiarity in instructional videos affects the learning effectiveness of college students (n=47). Two sets of instructional videos were produced that adopted video modelling to teach business etiquette. The forty-seven college students each viewed one of the two video sets with cast that are respectively familiar faces and unfamiliar faces to the students. The results showed that participants can learn effectively from both sets of videos. Further examination showed face familiarity significantly reduced learning effectiveness, only when the participants had full-time work experience; otherwise, face familiarity does not have effect on learning effectiveness. These findings were explained in accordance to the Cognitive Theory of Multimedia Learning and indicate that face familiarity may hinder learning effectiveness.
Massive Open Online Courses (MOOCs), which are open for anyone without limitations on time or loc... more Massive Open Online Courses (MOOCs), which are open for anyone without limitations on time or location, have attracted millions of registered online students. The large number of online courses available raises the question of how appropriate courses can be effectively recommended to interested learners. The recommendation system, widely used in various online applications, is a good solution for reducing decision complexity. In this paper, we propose the method of using attention-based convolutional neural networks (CNN) to obtain a user's profile, predict the user ratings, and recommend the top-n courses. First, we represent the learner behaviors and learning histories into feature vectors. The attention mechanism is then used to improve relevance estimation according to the differences between the estimation scores and the actual scores given by users to train the neural network. Finally, the trained model will recommend courses to learners. At the end of the paper, we introduce the framework of our system.
Answer ranking is one of essential steps in open domain question answering systems. The ranking o... more Answer ranking is one of essential steps in open domain question answering systems. The ranking of the retrieved answers directly affects user satisfaction. This paper proposes a new joint model for answer ranking by leveraging context semantic features, which balances both question-answer similarities and answer ranking scores. A publicly available dataset containing 40,000 Chinese questions and 369,919 corresponding answer passages from Sogou Lab is used for experiments. Evaluation on the joint model shows a Precison@1 of 72.6%, which outperforms the state-of-the-art baseline methods.
Answer ranking is one of essential steps in open domain question answering systems. The ranking o... more Answer ranking is one of essential steps in open domain question answering systems. The ranking of the retrieved answers directly affects user satisfaction. This paper proposes a new joint model for answer ranking by leveraging context semantic features, which balances both question-answer similarities and answer ranking scores. A publicly available dataset containing 40,000 Chinese questions and 369,919 corresponding answer passages from Sogou Lab is used for experiments. Evaluation on the joint model shows a Precison@1 of 72.6%, which outperforms the state-of-the-art baseline methods.
Computers and Education: Artificial Intelligence, 2021
Abstract Top-N personalized recommendation has been extensively studied in assisting learners in ... more Abstract Top-N personalized recommendation has been extensively studied in assisting learners in finding interesting courses in MOOCs. Although existing Top-N personalized recommendation methods have achieved comparable performance, these models have two major shortcomings. First, these models seldom learn an explicit representation of the structural relation of items. Second, most of these models typically obtain a user’s general preference and neglect the recency of items. This paper proposes a Top-N personalized Recommendation with Graph Neural Network (TP-GNN) in the Massive Open Online Course (MOOCs) as a solution to tackle this problem. We explore two different aggregate functions to deal with the user’s sequence neighbors and then use an attention mechanism to generate the final item representations. The experiments on a real-world course dataset demonstrated that TP-GNN could improve the performances. Furthermore, the system developed based on our method obtains positive feedback from the participants, which denotes that our method effectively predicts learners’ preferences and needs.
Face-to-face classes had been replaced by online classes from primary schools to universities aro... more Face-to-face classes had been replaced by online classes from primary schools to universities around the world due to COVID-19. In a 4th year computer science course, we forwent conventionally separated lectures and programming laboratories lasting 2 hours each and switched to alternating mini-lectures and student exercises several times in a 2-hour timeslot. Students reported that they learned better with improved motivation in this new class format. Students also indicated that they were mostly neutral about whether the face of professor was shown in online lecture.
Some non-native speakers struggle with lectures conducted in English. It may be hard for them to ... more Some non-native speakers struggle with lectures conducted in English. It may be hard for them to pick out key messages when reading books or lecture notes. Microsoft PowerPoint has an “add notes” feature. We propose to use it to attach lecture notes to individual slides instead of compiling all lecture notes in a conventional book form. Students learn key points from slides and detailed explanation from slide-based lecture notes. A suitable learning approach, deep or shallow, can be chosen according to personal learning goal on a slide-by-slide basis. Student-centered learning is empowered to a high level of granularity in a blended learning environment. Lecturers may update slide-based lecture notes more effectively and efficiently than separate sets of slides and lecture notes.
Outcome-based teaching and learning emphasizes the explicit declaration of learning outcomes whic... more Outcome-based teaching and learning emphasizes the explicit declaration of learning outcomes which identify the tasks students are expected to be able to perform after completing the course, and to what standard. OBTL also requires the teaching, learning and assessment activities to align with the stated learning outcomes. We interviewed fifteen university instructors about their experience of teaching outcome-based computer science
Communications in computer and information science, 2020
As an important educational form, online learning has attracted millions of registered learners, ... more As an important educational form, online learning has attracted millions of registered learners, and a huge number of courses are available online. However, it is challenging for learners to identify appropriate courses from a large course pool due to the difficulties of mapping complex learning needs to the high-level course semantics. Several studies in the field of Natural Language Processing (NLP) have recently gained promising performance in capturing the semantic information. In this study, we use these NLP techniques to understand the semantics of learning needs and courses. Specifically, we model users’ historical course records as word sentences using skip-gram with negative sampling to obtain course semantics. Furthermore, we introduce Laplacian Eigenmaps as the objective function and integrate the course social tags and course-user interaction as penalty factors to fine-tune the course vectors, especially the courses of different categories but similar contexts. The result verifies that the proposed method is effective for recommending suitable courses for users.
During the past five years, the development of various technologies and the maturing of related e... more During the past five years, the development of various technologies and the maturing of related education industries have never ceased, and the Hong Kong government continually invests into the comprehensive establishment of technology-enhanced education systems, especially those for English education. Here we then meet the point where the efforts of the past five years call for a panoramic view and an in-depth analysis of people's attitudes toward technology-enhanced language learning. This study investigated a total of 56 in-service English teachers' acceptance of technology-enhanced language learning and teaching. These teachers were from 56 different primary schools in Hong Kong. The results presented an overview of the current situation of e-learning in Hong Kong English education, showing insights into the future TELL design and development of related programs.
Students in a year 4 computing course were given a choice of learning activities using different ... more Students in a year 4 computing course were given a choice of learning activities using different media: live lectures, on-demand recorded lectures, and PowerPoint slides with/without transcripts. Students preferred slides with transcripts more than the other learning media because it enables them to concentrate better. However, students perform equally well regardless of their preferred choice of learning medium. We found that students who studied regularly week after week performed better than students who studied erratically. Our twofold recommendations are for courses designed to provide multiple learning media for students to choose and to encourage students to study regularly rather than in an erratic pattern.
In the past decade, it has become popular to use instructional videos for teaching and learning i... more In the past decade, it has become popular to use instructional videos for teaching and learning in online and blended learning environments. While researchers have studied how the presence of an instructor in an instructional video affects learning effectiveness, the influence of an instructor's visual familiarity toward students' learning is unclear. This experimental study explored how face familiarity in instructional videos affects the learning effectiveness of college students (n=47). Two sets of instructional videos were produced that adopted video modelling to teach business etiquette. The forty-seven college students each viewed one of the two video sets with cast that are respectively familiar faces and unfamiliar faces to the students. The results showed that participants can learn effectively from both sets of videos. Further examination showed face familiarity significantly reduced learning effectiveness, only when the participants had full-time work experience; otherwise, face familiarity does not have effect on learning effectiveness. These findings were explained in accordance to the Cognitive Theory of Multimedia Learning and indicate that face familiarity may hinder learning effectiveness.
Massive Open Online Courses (MOOCs), which are open for anyone without limitations on time or loc... more Massive Open Online Courses (MOOCs), which are open for anyone without limitations on time or location, have attracted millions of registered online students. The large number of online courses available raises the question of how appropriate courses can be effectively recommended to interested learners. The recommendation system, widely used in various online applications, is a good solution for reducing decision complexity. In this paper, we propose the method of using attention-based convolutional neural networks (CNN) to obtain a user's profile, predict the user ratings, and recommend the top-n courses. First, we represent the learner behaviors and learning histories into feature vectors. The attention mechanism is then used to improve relevance estimation according to the differences between the estimation scores and the actual scores given by users to train the neural network. Finally, the trained model will recommend courses to learners. At the end of the paper, we introduce the framework of our system.
Answer ranking is one of essential steps in open domain question answering systems. The ranking o... more Answer ranking is one of essential steps in open domain question answering systems. The ranking of the retrieved answers directly affects user satisfaction. This paper proposes a new joint model for answer ranking by leveraging context semantic features, which balances both question-answer similarities and answer ranking scores. A publicly available dataset containing 40,000 Chinese questions and 369,919 corresponding answer passages from Sogou Lab is used for experiments. Evaluation on the joint model shows a Precison@1 of 72.6%, which outperforms the state-of-the-art baseline methods.
Answer ranking is one of essential steps in open domain question answering systems. The ranking o... more Answer ranking is one of essential steps in open domain question answering systems. The ranking of the retrieved answers directly affects user satisfaction. This paper proposes a new joint model for answer ranking by leveraging context semantic features, which balances both question-answer similarities and answer ranking scores. A publicly available dataset containing 40,000 Chinese questions and 369,919 corresponding answer passages from Sogou Lab is used for experiments. Evaluation on the joint model shows a Precison@1 of 72.6%, which outperforms the state-of-the-art baseline methods.
Computers and Education: Artificial Intelligence, 2021
Abstract Top-N personalized recommendation has been extensively studied in assisting learners in ... more Abstract Top-N personalized recommendation has been extensively studied in assisting learners in finding interesting courses in MOOCs. Although existing Top-N personalized recommendation methods have achieved comparable performance, these models have two major shortcomings. First, these models seldom learn an explicit representation of the structural relation of items. Second, most of these models typically obtain a user’s general preference and neglect the recency of items. This paper proposes a Top-N personalized Recommendation with Graph Neural Network (TP-GNN) in the Massive Open Online Course (MOOCs) as a solution to tackle this problem. We explore two different aggregate functions to deal with the user’s sequence neighbors and then use an attention mechanism to generate the final item representations. The experiments on a real-world course dataset demonstrated that TP-GNN could improve the performances. Furthermore, the system developed based on our method obtains positive feedback from the participants, which denotes that our method effectively predicts learners’ preferences and needs.
Uploads
Papers by Oliver Au