April 6 2017, 16h – ENSSAT Lannion Bat. 3, room 214i
François Olivier-Martin (DCBrain): Visualisation de graphes de flux physiques. (résumé non demandé par l’équipe).
May 24 2017, 16h – ENSSAT Lannion Bat. 3, room 214i
Sara Al Hassad: Learning commonalities in RDF. Finding the commonalities between descriptions of data or knowledge is a foundational reasoning problem of Machine Learning introduced in the 70’s, which amounts to computing a least general generalization (lgg) of such descriptions. It has also started receiving consideration in Knowlegge Representation from the 90’s, and recently in the Semantic Web field. We revisit this problem in the popular Resource Description Framework (RDF) of W3C, where descriptions are RDF graphs, i.e., a mix of data and knowledge. Notably, and in contrast to the literature, our solution to this problem holds for the entire RDF standard, i.e., we do not restrict RDF graphs in any way (neither their structure nor their semantics based on RDF entailment, i.e., inference) and, further, our algorithms can compute lggs of small-to-huge RDF graphs.
October 12 2016, 16h – ENSSAT Lannion Bat. 3, room 214i
Henning Christiansen: An interactive installation for exhibition artist’s sketchbooks. We will explain the design of and experiences with an interactive oversize book installation that has been included in thematic exhibitions at prestigious art museums in Denmark, as well as in more experimental settings.
April 2 2015, 15h – ENSSAT Lannion Bat. 3, room 214i
Khadim Dramé: Interrogation de données structurées : une approche interactive à base de mots clés. Keyword-based search is a convenient way for users to express their information needs. Most of current information retrieval engines, such as Google, are based on keyword search. On the other hand, formal query languages (e.g. SPARQL), mainly used over structured data, are very ex- pressive but they are complex for end users. It requires knowledge of the language syntax but also the target data schema. Thus, keyword search is increasingly explored to simplify the querying of valuable structured data. In this work, we address this issue and propose a keyword-based approach over structured data. The principle is to generate candidate formal queries from keywords and then to propose an intuitive interface for guiding users to select the suitable one. For this, the latter are presented in a pseudo-natural language for helping casual users to understand the formal queries meaning.
February 26 2015, 15h – ENSSAT Lannion Bat. 3, room 214i
Aurélien Moreau: A Fuzzy Approach to the Explanation of Database Query Answers. Database query answers can be rather misleading: sometimes empty, sometimes plethoric, or without any form of ranking. Cooperative answers provide the user with complementary information such as data summaries, reformulations or more results. This paper describes an approach providing the user with more insight to better understand the results of their queries. Using clustering algorithms and fuzzy vocabularies, the idea is to form subgroups of answers based on their similar attributes. The originality of this work is that the data considered for characterizing each cluster of answers is not limited to the attributes used in the query. The objective is to enable the user to comprehend the structure of the results of their queries with labels from the natural language.
September 08 2014, 14h – ENSSAT Lannion Bat. 1, room 137C
Aurélien Moreau (Master 2 Research internship, UR1/Enssat – Irisa/Shaman, supervised by O. Pivert and G. Smits): Computation of Extended Answers to Relational Database Queries. This Master thesis is divided into two main parts: a state of the art and the work achieved during the internship. The bibliographic part highlights several aspects related to database browsing and recommendation systems. Assisting the user with keyword search or cooperative answers has been a recurring issue in the world of databases. Providing users with recommendations is of major importance in electronic commerce. Inspired by both recommendation systems and cooperative answers, the Shaman research team has outlined a new approach to compute similar answers. This approach based on fuzzy associations is at the heart of the Master’s degree internship. The main objectives include giving more insight as to why some answers are returned and automatically finding similarity criteria for a given object.
June 06 2014, 14h – ENSSAT Lannion Bat. 3, room 214i
Olfa Slama (Master 2 Research internship in Shaman, direction of L. Liétard and D. Rocacher): About alpha-cuts in bipolar and multipolar queries. The integration of user preferences into flexible queries gains more interest in the recent years. This is mainly due to its ability to satisfy user needs more closely. Recently, Bipolar and Multipolar queries have been introduced in order to deal with more complex preferences into queries such as “find hotel which is cheaper and if possible comfortable and if possible near the beach”. However, one of the most serious drawbacks of these queries is that the returned results may be enormous and not calibrated with respect to the user preferences. In such case, the user does not find what he/she seeks easily. To overcome this problem, we suggest an approach based on the concept of α-cuts in order to calibrate and improve the quality of the responses returned by Bipolar and Multipolar queries. This approach is then implemented and the experiment results show the promising results obtained with our approach with varying the α-value.
May 14 2014, 14h – ENSSAT Lannion Bat. 3, room 214i
Alexandra Roatis (INRIA Saclay OAK, Univ. Paris Sud): RDF Analytics: Lenses over Semantic Graphs (WWW’14). The development of Semantic Web (RDF) brings new requirements for data analytics tools and methods, going beyond querying to semantics-rich analytics through warehouse-style tools. In this work, we fully redesign, from the bottom up, core data analytics concepts and tools in the context of RDF data, leading to the first complete formal framework for warehouse-style RDF analytics. Notably, we define i) analytical schemas tailored to heterogeneous, semantics-rich RDF graph, ii) analytical queries which (beyond relational cubes) allow flexible querying of the data and the schema as well as powerful aggregation and iii) OLAP-style operations. Experiments on a fully-implemented platform demonstrate the practical interest of our approach.
April 24 2014, 11h – ENSSAT Lannion Bat. 3, room 214i
Stamatis Zampetakis (INRIA Saclay OAK, Univ. Paris Sud): CliqueSquare – Efficient RDF Query Processing in MapReduce-like Systems. The RDF data model has gained a lot of popularity from both industry and academia. Applications are creating, publishing and using very large volumes of RDF data in several contexts, e.g., in the Linked Data movement. Managing very large volumes of RDF data is challenging due to the sheer size of the data and the particularity of RDF queries, which quite often involves many self-joins. We propose CliqueSquare, a scalable RDF data management system for the efficient RDF query evaluation in MapReduce-like systems. CliqueSquare provides a novel RDF data partitioning scheme that aims at reducing the amount of data network transfers network during query evaluation. CliqueSquare uses a novel query optimization algorithm, based on graphs and cliques, which enables high degrees of intra-query parallelism. As a result, CliqueSquare achieves very low query runtimes. We implement CliqueSquare on top of Hadoop and experimentally compare it with the state-of-the-art Hadoop-based RDF store. The results show an improvement of up to more than one order of magnitude in terms of query performance and network traffic.
April 24 2014, 10h – ENSSAT Lannion Bat. 3, room 214i
Ioana Manolescu (INRIA Saclay OAK, Univ. Paris Sud): Presentation of the OAK team (INRIA Saclay Île-de-France).