BI and Semantic Web Colloquim

Session Descriptions

Some pictures of the event

Monday 08:30 - 09:00 Organizers/Opening

A meeting of the Colloquium Organizers and local chairs up before the start of the coloquim.

Why this Colloquim? (slide) (Dr. Ghislain Atemezing)

Monday 09:00 - 10:00 Keynote Axel-Ngonga (Amphi du Bloc Pédagogique de la Faculté des Sciences)

• Speaker: Prof.Dr. Axel-Cyrille Ngonga

Title: Trends in Big Data Analytics: Applications in Research and Industry Projects

(slide)

Bio: Prof. Dr. Axel-Cyrille Ngonga Ngomo is a full professor of Data Science at Paderborn University. He obtained his PhD in Computer Science at Leipzig University in 2009. He then led the Agile Knowledge and Semantic Web Research group, where he focused on knowledge extraction, data storage, federated query processing and data integration for knowledge graphs. His research was further extended to structured machine learning with his appointment to a full professorship in 2017. Axel now leads the DICE research group at Paderborn University, which develops scalable web-scale algorithms to support all steps of the life cycle of knowledge graphs. His research has led to more than 200 peer-reviewed publications and more than 25 international research awards including Best Research Papers at top-tier conferences and the 2016 Next Einstein Forum award. Axel has led the development of or developed more than 100 open-source frameworks for knowledge graphs, which are available at dice-research . He currently leads the European Training Network KnowGraphs on knowledge graphs at scale. He is also the scientific lead of half a dozen projects on topics ranking from fact checking to machine learning. In his free time, he enjoys a good jam session on guitar, bass and sometimes even drums.

Monday 10:00 - 11:00 Keynote Maria Keet (Amphi du Bloc Pédagogique de la Faculté des Sciences)

• Speaker: Dr. Maria Keet

Title: Ontology verbalization for African languages

Abstract: Ontology verbalization involves generating natural language text from structured representations of knowledge, such as ontologies in OWL. This has multiple uses for business, such as for subject domain knowledge acquisition for application development, electronic health records management, and advanced data analysis with conceptual queries. However, neither OWL nor the common Semantic Web technologies infrastructure cater for easy tweaking to accommodate the core characteristics of African Languages for ontology verbalization. To achieve this nonetheless, we have been developing novel algorithms, models, and frameworks. While the test case language is mostly isiZulu, bootstrappability and generalizability of the algorithms to other African languages have been demonstrated, and the models can serve also other languages. This talk aims to provide an overview of the principal theoretical and engineering factors and approaches for successful ontology verbalization for the Niger-Congo family of African languages.

Bio:C. Maria Keet (PhD, MSc, MA, BSc(hons)) is an Associate Professor with the Department of Computer Science, University of Cape Town. She focuses on knowledge engineering with ontologies, conceptual data modelling, and their interaction with natural language, which has resulted in over 100 peer-reviewed publications at venues including K-CAP, EKAW, KR, FOIS, ESWC, ER, CIKM, Applied Ontology, Data & Knowledge Engineering, and the Journal of Web Semantics. She has been PI and researcher in several internationally and nationally funded projects. She wrote the first textbook on ontology engineering for computer scientists. Before her employment at UCT, Maria was a Senior lecturer at the School of Computer Science, University of KwaZulu-Natal, South Africa. Before that, she was a non-tenured Assistant Professor at the KRDB Research Centre, Free University of Bozen-Bolzano, Italy, where she also obtained her PhD in Computer Science (2008), following a BSc(honours) 1st class in IT & Computing from the Open University UK in 2004, and 3.5 years work experience as systems engineer in the IT industry. She also obtained an MSc in Food Science from Wageningen University, the Netherlands (1998) and an MA 1st class in Peace & Development Studies from the University of Limerick, Ireland (2003).

Monday 11:00 - 11:30 Coffee Break (Amphi du Bloc Pédagogique de la Faculté des Sciences)

Monday 11:30 - 12:15 Keynote Michel Kana Keet (Amphi du Bloc Pédagogique de la Faculté des Sciences)

• Speaker: Dr. Michel Kana

Title: Landscape of analytics management solutions in Social Marketing

Abstract: Let's explore data science and social media for a moment. When were you on Facebook, Instagram, LinkedIn, or Twitter for the last time? 5 minutes ago? Internet users are now spending an average of 2 hours and 22 minutes per day on social media and messaging platforms. 2020 will be the year that social commerce takes off, as more and more customers make purchases directly from social media platforms. 60% of all total ad spend is allocated to the Facebook News Feed. The Instagram feed comes in a distant second at 20%, followed by Stories at 10%, and the rest of the top 5 – Facebook suggested video, and Facebook instream video – combine to 10%. Despite of Ads, customers increasingly rely on endorsements and product placements from influencers, people and organizations who possess a purported expert level of knowledge and social influence in their fields. We will not only explore how Socialbakers helps marketers with its leading SaaS suite. We will also dive into an illustrative dataset of influencers by performing hands-on data exploratory analysis and machine learning with Python. Together we will get our hands dirty and our heads clean. Leading brands like Estee Lauder, Boss and Burberry are investing in influencer marketing as a key driver of their success on social media. You can also be part of this revolution.

Bio: Dynamic, accomplished, multidisciplinary Data Science Product Owner at Socialbakers highly regarded for guiding software engineering efforts and implementing machine learning techniques to solve complex problems within large distributed systems. Out-of-the-box thinker with extensive knowledge of mathematical modeling, neural networks, deep learning and statistical concepts, combined with an extensive background in dealing with large datasets and deploying real-time algorithms for Fortune 500 companies. Known for ability to link neuroscience and machine learning through academic engagements, research experience, and lecturer experience in modeling/simulation of biological processes. As a Prince 2 Certified, recognized for effectively guiding and supporting critical projects, understanding stakeholder requirements and delivering on all goals. Accepted as a strong communicator who ensures continued coordination with diverse individuals. Demonstrated Startup founding spirit at Digintu.tech, social entrepreneurship at DiGIT and volunteer work concerning tech charity for kids at Karewa and Yemba.net.

Monday 12:15- 13:00 Keynote Djofang Dieudonné (Amphi du Bloc Pédagogique de la Faculté des Sciences)

• Speaker: Msc. Djofang Dieudonné

Title: La lutte contre la fraude au moyen du datamining. (slide)

Abstract: La lutte contre la fraude apparait aujourd’hui comme un réel enjeu pour les entreprises. Afin de détecter les comportements frauduleux, les entreprises utilisent des règles métiers établies par des experts qui ne s’adaptent pas toujours à l’ingéniosité et à la créativité des fraudeurs. Par la mise en place au moyen d’analyse avancées, de modèles prédictifs sophistiqués à associer aux règles métiers existantes, le data mining se présente comme un outil incontournable de lutte contre la fraude. Les entreprises ressentent également de plus en plus le besoin d’appliquer les modèles prédictifs sur des données en temps réel. La prédiction en temps réel trouve son utilité dans différents domaines :

• Dans le domaine industriel, les capteurs génèrent des informations qui peuvent être exploitées en temps réel, par exemple pour anticiper une panne matérielle.
• Dans le domaine de la télécommunication, notamment pour détecter en temps réel les intrusions et limiter les trafics illicites afin de renforcer la sécurité informatique; identifier les fraudes à la carte SIM telles que le clonage de carte.
• Dans le domaine du e-marketing, pour les propositions instantanées des produits sur un site internet suite à l’estimation du score d’appétence.
• Dans le domaine monétique bancaire où les données sont scorées lors de la transaction électronique pour identifier des suspicions de fraude.
Ce travail a pour but de présenter la méthodologie de détection de la fraude à l’aide du datamining et d’illustrer l’application d’un modèle prédictif sur un flux de données en temps réel. Il s’articule donc en deux parties :
• une première partie qui illustre la méthodologie de détection de la fraude à l’aide du datamining, en s’appuyant sur un jeu de données issu du domaine de l’assurance,
• et une seconde partie qui illustre comment appliquer un modèle prédictif sur un flux de données en temps réel. Ce dernier cas s’appuie sur un flux de données monétiques bancaires (transactions par cartes bancaires).
Dans la première partie, après une présentation des différents types de fraude dans le domaine de l’assurance, nous décrivons les conditions favorables au mécanisme de détection de la fraude, les différentes techniques et algorithmes existants pour identifier les comportements frauduleux. Les logiciels SAS Base, SAS Enterprise Guide sont ensuite utilisés dans la préparation et l’analyse d’un jeu de données issu du domaine de l’assurance automobile. Le logiciel SAS Enterprise Miner a servi à construire le processus de modélisation et à sélectionner un modèle optimal de détection des fraudeurs. Dans la seconde partie, nous nous intéressons à l’application d’un modèle de prédiction sur un flux de données en temps réel. Une telle utilisation du modèle nécessite un environnement opérationnel adéquat. Pour cela, nous utilisons le logiciel de modélisation KXEN et les technologies PHP, JAvascript, SQL SERVER pour construire un environnement de démonstration.

Bio: Dieudonné DJOFANG est consultant sénior en BI. Après un diplôme de DEA en sciences informatiques obtenu à l’université de Yaoundé1 en 2005, Dieudonné DJOFANG a exercé dans un groupe bancaire sous-régional Africain ayant pour siège Yaoundé. En 2012, il a quitté son poste de coordonnateur des projets monétiques pour une spécialisation en Informatique Appliquée à la Décision Bancaire et Actuarielle à l’école Supérieure de Télécommunication de Bretagne (ENSTB) en France. Depuis 2013, après obtention d’un diplôme de mastère spécialisé IADBA, il exerce aujourd’hui en tant que consultant data (data Engineer) auprès des entreprises financières du CAC40 (banques et assurances). Il justifie d’une forte expertise sur la solution décisionnelle SAS avec plusieurs certifications professionnelles à son actif.

Monday 13:00 - 14:00 Lunch (Dept Info)

Monday 14:00- 14:30 Keynote Ghislain Atemezing (Amphi du Bloc Pédagogique de la Faculté des Sciences)

• Speaker: Dr. Ghislain Atemezing

Title: Big Data and Vocabularies for Actionnable Intelligence

Abstract: What do a consumer goods manufacturer and a credit insurance group have in common? Both are subject to a variety of risks which, if not detected, may dramatically impact their operations and bottom lines. Delve into the challenges of putting together a semantic, technology-based business solution that monitors and reacts to a large amount of consumer feedback in real time, providing insights on consumer product quality. This talks presents the approach assisting credit risk analysts in the early detection of signals and events affecting companies’ solvency to anticipate default risks of targeted companies. Walk through this journey to solve real-world problems with business intelligence solutions based on semantic data and technologies.

Bio: Dr. Ghislain Atemezing obtained a PhD in computer science from Telecom ParisTech (France) and is a recognized expert in ontology web language, semantic technologies and triple store systems. He joined Mondeca (a software company in semantics applications based in Paris, France) in 2015. Ghislain heads up Mondeca's knowledge engineering practice (graph database expertise, data modeling and transformation, LOD querying and scraping) and helps enrich our clients’ data sets with linked open data. Ghislain is also in charge of promoting, supporting and curating the Linked Open Vocabularies initiative, the reusable linked vocabularies ecosystem shared by ontology experts around the globe. Main work on projects include ontology modeling, RDF stores benchmarking, text analytics, data model validation and fine tuning, linked open data querying, scraping and clean up (e.g. from such as Geonames, DBpedia, Wikidata, etc.). Ghislain also give lectures in Semantic Web to Master 2 students at CNAM and Paris Descartes University.

Monday 14:30- 15:30 Keynote Gayo Diallo (Amphi du Bloc Pédagogique de la Faculté des Sciences)

• Speaker: Dr. Gayo Diallo (HDR)

Title: Big Data and Semantic-based Digital Public Health: focus on Low and Middle Income Countries. (slide)

Abstract: Our daily life is nowadays tremendously affected by the digital revolution. Information and Communications Technologies (ICT) are used for improving human health and services both at individual and population level. In particular, Artificial Intelligence (AI) methods and tools are progressively changing the way we are tackling healthcare related issues. In this talk I will address the issue of the abundance of heterogeneous data and knowledge in the context of Public Healths and the challenges that it raises. I will focus in particular on some multidisciplinary works related to bringing semantic technologies and AI based approaches that we are conducting to enable digital public health: from drug discovery and tackling pharmacovigilance issues using online forums and social networks through public health decision enabling thanks to the analysis of large call data records (CDR).

Bio: HDR and PhD in Computer Sciences, is a member of the Bordeaux Laboratory of Computer Science and the Group Leader of the Computer Sciences Applied to Health research group (ERIAS) of the Bordeaux Population Health research center, INSERM 1219, a team dedicated to the design and development of methods and tools for the semantic integration of healthcare related data, in particular, for facilitating their secondary use in the context of Public Health issues. He joined University of Bordeaux, Bordeaux School of Public Health (ISPED), in 2009 after being a research assistant at City University of London (UK) and PostDoc researcher at the Laboratory of Applied Computer Sciences (LISI/ENSMA) Futuroscope Poitiers (France). He graduated from University of Grenoble Joseph Fourier in 2006 (PhD in Computer Science). His research interests include AI based approach for healthcare data and knowledge management and ICT for Development with a particular focus on the healthcare sector. He participated in various EU funded projects and authored or co-authored more than 50 peers reviewed papers. He is winner of the practical application prize of the 2015 edition of the Orange Data for Development Big Data Challenge (D4D) and the SAMPO France-Finland cooperation program in 2015 as well. Web site: Gayodiallo.org

Monday 15:30- 16:15 Keynote Gaoussou Camara (Amphi du Bloc Pédagogique de la Faculté des Sciences)

• Speaker: Dr. Gaoussou Camara

Title: Ontology-based data annotation and integration for disease surveillance. (slide)

Abstract: In this talk we will present a feedback on the use of ontologies in disease surveillance through the VEMIS and MABO projects. The aim of the Epidemiological Surveillance of Infectious Diseases in Senegal (VEMIS) project is to design a health surveillance system based on ontologies for a composition of numerical and qualitative simulation models with different formalisms. The project Management and Access of Biomedical Knowledge through Ontology Network (MABO) focuses on the management of biomedical knowledge and their access based on an ontology network. We will also talk about ontology learning, based on hidden Markov chains, from the Java source code of the epidemiological surveillance platform for tuberculosis in Cameroon (EPICAM).

Bio: Dr. Gaoussou Camara is an Associated Professor in Computer Science in the Department of Mathematics of the Applied Sciences and ICT faculty of Alioune Diop University of Bambey (UADB). He is leading the Interdisciplinary Research Team in Medical Informatics and Information and Communication Technology in Education (IMTICE). His research focuses on the management of biomedical data and knowledge. On the one hand, he co-founded the SIMENS project which aims at designing an electronic health record (EHR) and a hospital information system (HIS) in Senegal. This project also proposes a set of decision-making tools to the health and political authorities. On the other hand, he is interested in domain and process ontology engineering for supporting infectious disease surveillance systems. The purpose is to demonstrate that the use of these ontologies could support and improve the communication between multidisciplinary actors, the integration of heterogeneous data coming from different sources, the composition of the numerical and qualitative simulation models of different formalisms, and data collection.

Monday 16:45 - 17:00 Coffee Break (Amphi du Bloc Pédagogique de la Faculté des Sciences)

Monday 17:00- 17:30 Keynote Pierre Lotis Nankep (Amphi du Bloc Pédagogique de la Faculté des Sciences)

• Speaker: M. Pierre Lotis Nankep

Title: Application centrée sur les données dans le système des marchés publics au Cameroun. (slide)

Abstract: L’exploration de l'applicabilité du paradigme du business intelligence dans la gouvernance publique est la problématique mise en exergue dans cette communication. En entreprise, afin de soutenir ou d’accroitre la performance économique, on fait recours aux technologies numériques pour une analyse proactive et en temps réel des données générées ou du marchés. Dans l’administration ou le service public, les enjeux majeurs concernent l’amélioration permanente de la qualité de la dépense publique. Ce qui suppose une maîtrise de la masse sans cesse croissante des données générées par diverses sources. La dépense publique, qui est mise en œuvre à travers les marches publics, nécessite un système cohérent, efficace et efficient de collecte, de stockage, de traitement, de mesure, d’évaluation de prise de décision et de planification ou projection. L’essentiel de cette communication est axé sur l’application Pridesoft, qui utilise exploite les flux de données générées par le système des marchés publics camerounais afin d’améliorer les politiques publiques en matière de dépense publique.

Bio: Pierre Lotis NANKEP est titulaire d'un DEA en Informatique obtenu en 2001 à l'Université de Yaoundé I au Cameroun. Il est expert en ingénierie logicielle pour laquelle il compte à son actif plusieurs applications informatiques en exploitation dans les entreprises privées et administrations publiques au Cameroun. Il justifie d’une expertise sur les questions de cybersécurité en général et de cryptographie en particulier (parcours CISSP accompli) avec plusieurs certifications professionnelles sur les système Linux. Il bénéficie également d’une solide expérience en matière de conduite de projet (parcours PMP accompli). Il Totalise plus de 15 années d’expériences donc 9 à l'Agence Nationale des Technologies de l'Information et de la Communication (ANTIC) au Cameroun et 5 à l’Agence de Régulation des Marchés Publics (ARMP) où il est le Directeur des Systèmes d’Information.

E-mail: lnankep@gmail.com, pierre.nankep@armp.cm

Monday 17:30- 18:00 Keynote Christelle Yemdji (Amphi du Bloc Pédagogique de la Faculté des Sciences)

• Speaker: Dr Christelle Yemdji Tchassi, Software Engineer, Renault, France

Title: Autonomous Vehicles and Data Mining. (slide)

Bio: LinkedIn

Monday 18:00 - 18:30 Wrap Up / End (Amphi du Bloc Pédagogique de la Faculté des Sciences)

• Go for a beer somewhere in Yaounde (ideas?)
• Dinner with the speakers and organizers (venue - TBC)