Arabic has a complex structure, which makes it difficult to apply natural language processing (NLP). Much research on Arabic NLP (ANLP) does exist; however, it is not as mature as that of other languages. Finding Arabic roots is an important step toward conducting effective research on most of ANLP applications. The authors have studied and compared six root-finding algorithms with success rates of over 90%. All algorithms of this study did not use the same testing corpus and/or benchmarking measures. They unified the testing process by implementing their own algorithm descriptions and building a corpus out of 3823 triliteral roots, applying 73 triliteral patterns, and with 18 affixes, producing around 27.6 million words. They tested the algorithms with the generated corpus and have obtained interesting results; they offer to share the corpus freely for benchmarking and ANLP research.

MOOCs offer free access to educational materials, leading to large numbers of students registered in MOOCs courses. The MOOCs forums allow students to post comments and ask questions; due to the number of students, however, the course facilitators are not able to provide feedback in a timely manner. To address this problem, we identify content-knowledge related posts using a course domain ontology and provide students with timely informative automatic feedback. Moreover, we provide facilitators with feedback of students posts, such as frequent topics students ask about. Experimental results from one of the courses offered by Coursera 3 show the potential of our approach in creating a responsive learning environment.

This paper investigates the impact of using different indexing approaches (full-word, stem, and root) when classifying Arabic text. In this study, the naïve Bayes classifier is used to construct the multinomial classification models and is evaluated using stratified k-fold cross-validation ( k ranges from 2 to 10). It is also uses a corpus that consists of 1000 normalized Arabic documents. The results of one experiment in this study show that significant accuracy improvements have occurred when the full-word form is used in most k-folds. Further experiments show that the classifier has achieved the highest accuracy in the eight-fold by using 7/8–1/8 train–test ratio, despite the indexing approach being used. The overall results of this study show that the classifier has achieved the maximum micro-average accuracy 99.36%, either by using the full-word form or the stem form. This proves that the stem is a better choice to use when classifying Arabic text, because it makes the corpus dataset smaller and this will enhance both the processing time and storage utilization, and achieve the highest level of accuracy.

Evidence based on ongoing published research shows that timetabling has been a challenge for over two decades. There is a growing need in higher education for a learner-centered solution focused on individual preferences. In the authors' earlier published work, students' group assessment information was mined to determine individualized achievements and predict future performance. In this paper, they extend the work to present a solution that uses students' individualized achievements, expected future performance, and historical registration records to discover students' registration timing patterns, as well as the most appropriate courses for registration. Such information is then processed to build the most suitable timetable for each student in the following semester. Faculty members' time preferences are also predicted based on historical teaching time patterns and course teaching preferences. The authors propose a modified frequent pattern (FP)-tree algorithm to process the predicted information. This results in clustering students to solve the timetable problem based on the predicted courses for registration. Then, it divides the timetable problem into subproblems for resolution. This ensures that time will not conflict within the generated timetables while satisfying both the hard and soft constraints. Both students' and faculty members timetabling preferences are met (88.8% and 85%).

Massive Open Online Course (MOOC) systems have recently received significant recognition and are increasingly attracting the attention of education providers and educational researchers. MOOCs are neither precisely defined nor sufficiently researched in terms of their properties and usage. The large number of students enrolled in these courses can lead to insufficient feedback given to the students. A stream of student posts to courses' forums makes the problem even more difficult. Students'-MOOCs' interactions can be exploited using text mining techniques to enhance learning and personalise the learners' experience. In this paper, the open issues in MOOCs are outlined. Text mining and streaming text mining techniques which can contribute to the success of these systems are reviewed and some open issues in MOOC systems are addressed. Finally, our vision of an intelligent personalised MOOC feedback management system that we term iMOOC is outlined.