flKeynote Speakers

First keynote speaker

Dr. Mohamad MRAYATI
Dr. Mohamad MRAYATI


  • Doctorat d’Etat es-Sciene Physique, Institut National Polytechnique de Grenoble, France (1976).
  • Diplôme d’Etude Approfondie (DEA) en Electronique, Université Scientifique et Medicale de Grenoble, et Ecole (ENSERG) France (1971)
  • Bachelor of Electrical Engineering, American University of Beirut (1968)


  • Expert on Science and Technology for sustainable Development. The German technical cooperation GIZ, since 1/6/2010, till now.
  • Regional Adviser at the United Nations – ESCWA (1997 until 1-4-2006), Then with UN-DESA as senior advisor on S&T for sustainable development. Ministry of Economy and Planning, KSA, from 2006 till 30/5/2010.
  • Director and Professor at the Higher Institute of Applied Science and Technology (HIAST) in Damascus (1990-1997).
  • Director of the Institute of Electronics Research at SSRC – Damascus until 1990.

Research interests and contributions

Work on establishing policies and strategies in Knowledge Based Economy, Science and Technology; sustainable development, mechanisms to link education, R&D with production and services activities. He is also interested in technology transfer operations and in Information and Communication projects

Has numerous researches in signal processing; in information technology and the Arabic language; in language technology (speech and text); in e-commerce, and in standardization.

Representative Publications

  • Co-author of three published scientific books in Arabic in the fields of the Arabic language, and computers.

  • Participated in writing chapters in five published books in English and two in Arabic.

  • Published more than 30 scientific articles in specialized refereed journals.

  • Presented more than 150 articles in refereed scientific conferences.

  • Submited over 60 technical advisory reports since he joint UN- ESCWA

  • Has a registered patent in Europe and the United States.



Second keynote speaker

Dr. Mohamed Attia Mohamed El Araby Ahmed
Dr. Mohamed Attia Mohamed El Araby Ahmed

Lexical Semantic Coherence as a Quantitative Metric of the Meaningfulness of Degraded Arabic Text in NLP Applications

  Abstract of speech


In this lecture we aim at presenting two main ideas; the first is to formally quantify a metric of the lexical semantic coherence of a given piece of text so that its value is consistent with its meaningfulness as perceived by native speakers. Lexical semantic coherence in this context is to be automatically computed through some form of weighted pairwise correlation of the lexical semantic similarity across all the words in the given piece of text. Lexical semantic similarity is in turn to be computed through some inverse formula in terms of an even lower-level metric that we call the lexical semantic distance between two given words. Our view is to calculate the lexical semantic distance according to the relative semantic locations and relations of the two words in a well-built ontology of the language.

While the concept of text coherence has been proposed several decades ago as a subjective classical linguistic one, we aim at projecting this concept to lexical semantic coherence and rigorously defining it as a general quantitative metric of text correctness. This metric is to be deterministically computed in reliance on some well-built lexical semantic resource.

Deterministically computing the semantic coherence of a given piece of text is of a great value as a source of judgment on its correctness independent from other corresponding stochastic metrics; e.g. probabilities based on statistical language models (SLM). According to the well-established data fusion theorems, classification based on multiple independent sources of data is better than (or at least equal to) the best achievable classification based on any single one of these sources.

The second main idea is then to capitalize on this independence by merging the metric of lexical semantic coherence with other metrics based on statistical language models for a better judgment on the correctness of degraded text; esp. that generated by automatic NLP/HLT recognition systems such as ASR, OCR, and MT. Specifically, we formulate the cost function that guides the lattice-search algorithms - like A* - in such systems, from two components; a stochastic one based on SLM probabilities and a deterministic one of our lexical semantic coherence metric.


  Brief CV

PhD of Electronics and Electrical Communications


Starting from July 1st, 2014 Dr. Attia serves as:

☆ Computer Scientist for the project of Doha Historical Dictionary of the Arabic Language.

He has been serving over the past years as a:

S/W Engineering Architect & Consultant, Luxor Technology, Oakville-Ontario-Canada

Visiting professor of CS and IT at different academic institutes in MENA region

Independent Researcher and Project Manager (Canada, Egypt, MENA region)

Consultant of Human Language Technologies, RDI www.RDI-eg.com, Giza-Egypt

He has previously led RDI's team of researchers and software engineers from 2000 to 2006 that succeeded to develop leading Arabic NLP technologies such as:

  • Hafss© the interactive self-learning assistant of spoken Arabic language and Tajweed,

  • ArabMorpho© the morphological analyzer and disambiguator of Arabic text,

  • ArabDiac© the automatic diacritizer of Arabic text,

  • Fassieh© the visual interactive tool for the morphological, diacritical, and lexical semantic annotation of Arabic text,

  • CleverPage© the omnifont OCR of printed Arabic text,

  • etc.

He has also led RDI’s team building the large-scale Arabic ontology based on semantic fields from 2005 to 2009 in the course of Data Mining and Computer Modeling Center of Excellence; http://dmcm.cu.edu.eg/ established by the Egyptian Ministry of Communications & IT (MCIT).

R&D Interests and Areas of Expertise

Dr. Attia has 19 years of industrial and academic R-to-D experience in the following areas:

  • Artificial intelligence,

  • Machine learning,

  • Classical and computational linguistics,

  • Written & spoken natural language processing, and human language technologies,

  • Signal processing,

  • Image processing,

  • Pattern recognition,

  • Software engineering,

  • Technical writing in English and in Arabic.

He has been a major figure in multinational projects and consortia of Arabic NLP/HLT such as the EC’s FP6 and FP7 NEMLAR and MEDAR www.MEDAR.info. He has also been the person in charge of the local organization of the NEMLAR’s Int’l conference on Arabic HLT http://www.ELDA.org/NEMLAR-conf, Cairo - Egypt, Sept. 2004, and the 2nd Int’l Conference on Arabic Language Resources and Tools www.medar.info/conference_all/2009/index.php Cairo - Egypt, Apr. 2009.

His publications appear as research papers in ISI-indexed journals and first-class conferences, as well as book chapters and books produced by major scientific publishers like Springer and Elsevier. Over the last couple of years, he has also reviewed several tens of manuscripts for ISI-indexed journals published by Springer, Elsevier, and ACM.