One of the important things I learned from my long teaching experience is to keep in mind that most students are not wrong, but rather they are different. I should then provide all necessary efforts in facilitating complex design and analysis tasks to all my students and create ways that objectively measure students’ learning and not reduce learning the algorithms concepts and complexity analysis to mere memorization. I have to assist my students to develop critical thinking skills that will enable them to a better understanding of problem-solving using the appropriate algorithm. I should always be available to reply to the students’ questions and be the ice breaker instrument in their hands.

I like teaching programming related courses like programming concepts, advanced programming, object oriented programming, data structure and algorithms. I teach also courses on discrete mathematics, logic, expert systems and mathematics of computing.

This semester I am teaching the following courses:

CMPS 323 Design and Analysis of Algorithm

Catalog Description: Analysis, design, and efficiency of algorithms illustrated by a comprehensive exposure to fundamental algorithms and various adopted techniques to solve different types of problems. Analysis of sorting, searching, and other algorithms; designing algorithms using techniques for problem-solving such as greedy methods, divide-and-conquer, backtracking, dynamic programming, and branch-and-bound techniques; complexity of algorithms.

CMPS 151 Programming Concepts

Exposure to problem solving techniques and operations on data using the fundamental components of a programming language. Problem solving techniques and presentations; motivations to programming languages and program execution; fundamental components of a programming language including simple and structured data representation; mathematical and logical operations; input/output, control and loop structures; functions; recursion; memory referencing; and simple file processing.

Senior Project: QUShopHere Portal

The main objective of this project is to develop and deploy a complete platform for online transactions at Qatar University. It consists, of building, installing and testing a web portal to buy and sells goods at the University. Staff and students at Qatar University can sell or buy items including cars, furniture, perfumes, clothes or other items online. They would be able to create an account using their QU email and ID to login into the portal and get identified. They can post the advertisement with images and multimedia and set the requested price. Interested buyer mainly from Qatar University can contact the sellers either through a discussion forum, email, SMS or simply by phone. The sellers can update their information which includes adding new images, reducing the prices, or any other useful information. The information can be either controlled by a moderator or simply by the sellers and buyers as the portal is currently dedicated to QU people only. The buyers can also launch an auction on the merchandise or stuff they want to sell quickly. The buyers can also ask the moderator to take photos and generate multimedia for the item they want to sell in a professional way against some fees. The portal can be deployed in QU server or in a specific domain that we can hire/subscribe with annual subscription.

PhD thesis-1: Arabic Digital Media Intelligence

There has been vast proliferation in Digital Arabic Content (DAC) production. However, there have been low advances in developing systems dedicated to the semantic management of DAC. While previous technologies have been developed attempting to tackle the semantic analysis of digital content, there exists no global approach or solution to the low level analysis of Arabic media. The vast majority of digital Arabic AV content found on social media platforms, information databases, professional archives and such, is available in “raw” form, uncategorized and unclassified. This type of “raw” DAC data is steadily increasing in availability and number day by day. Large portions of this data remain unused and are constantly replaced or edited with new data. The reason that this digital material is not being reused and that large amounts of content are unaccounted for, is that the production chain of such data does not include the vital step of content structuring. The lack of audio-visual search and retrieval tools for capturing unannotated digital media assets is the cause of the lack of reuse of this data. To address these and several other shortcomings of current multimedia systems, the research community has launched initiatives such as the MPEG-7 and MPEG-21 international standards to facilitate content-based representation and description of audio-visual data. However, the problem of inferring the semantics of the content is still an open research issue. Moreover, these standards are not language-oriented and do not accommodate linguistic, social and cultural specifications of Arabic digital media scenes. In this context, the objective of this study is to design, implement and assess a multimodal system for automated classification of Arabic videos. The proposed system will target both raw and general purpose Arabic content found on a variety of platforms on the Web involving multimodal video processing. Raw material has some distinct constraints that must be considered during development: camera settings, shot-boundary detection, soundtrack irrelevance, redundancy detection, irrelevant annotations, and isolated fragments. The proposed system will be composed of the following components: utilization of Arabic Named Entity Recognition (NER) for text-based video classification, combined with audio-based analysis to extract patterns for domain-based video classification. The classification domain targeted is “news” videos, pertaining to “shooting” and “explosion”. NER is a task that locates, extracts and automatically classifies named entities into predefined classes in unstructured texts. It covers proper names, temporal expressions and numerical expressions. Proper names are classified into three main groups: persons, locations and organizations. The majority of NER studies have been focused on the English language, as it is the internationally dominant language, while research on other languages for the NER task has been limited.  Audio approaches require fewer computational resources than that of visual methods. When features are stored, they also require less space. Another advantage is that segmented audio clips tend be very short (average 1-2 seconds), so the processing of the audio clips would be easier. Audio features can lead to three layers of audio understanding: low-level acoustics, such as the average frequency for a frame, midlevel sound objects, such as the audio signature of the sound a ball makes while bouncing, and high-level scene classes, such as background music playing in certain types of video scenes.

PhD thesis-2: Educational Ontology Based Question Answering

Master thesis-1: Helping Qatar University students with learning difficulties understand PDF Arabic-based text

Students with special needs are facing many difficulties in their life especially in learning field. As known, blind people can’t use basic computer software because they can’t issue commands by keyboard/mouse nor display the output on the screen. Since they are part of the community and due to digital evolution and inflation of documents over the Internet, it becomes necessary to create programs that help them to keep up with this development and overcome their limitation. For that, Text to Speech (TTS) and Speech to Text (STT) techniques are discovered to convert written text to audible one and vice-versa. Those tools must take editable text to read, like files written in Word/Excel/Notepad. Converting automatically Arabic PDF files to editable files (i.e., MS-Word) is a big challenge and the current available software produce inaccurate results which requires manual auditing. In fact, the best available software gives loosely 60% of accuracy and manual intervention are always required. This task requires huge efforts from editors and it is not possible to convert all the available PDF Arabic-based text as their number is growing continuously. Hence, the objective of this work is to develop a system that can increase the accuracy of converting Arabic PDF documents to editable texts where it can be read loudly by the reading software. In the context of helping blind students in Qatar university in their learning process, we need to scan books or get electronic PDF books written in Arabic language and then convert them to MS-Word format in order to use TTS applications (i.e., automatic reading tools). Usually this conversion results in corrupted files, which requires to read the whole document from specialists and update the texts manually. Imagine when you have a big book with 1000+ pages, it will take huge efforts, time and resources to accomplish the correction task. Therefore, trying to optimize current OCR software to support Arabic language plus detecting corrupted words is the core of this master project. We will study the existing software (i.e. IRIS software) and develop techniques to improve the accuracy. This may result in developing a comprehensive system that uses a classifier to recognize Arabic words using some features extracted from the trained text, like histogram. We will develop an easy to use interface that can be used by the potential users of the system. We will test the proposed system on a collection of PDF documents on different domain.