ENIRAF · Enhanced information retrieval and filtering for analytical systems

Final Activity Report Summary - ENIRAF (Enhanced information retrieval and filtering for analytical systems)

ENIRAF aimed at supplementing internal, structured data of the enterprises with external, unstructured information. The project concerned a variety of topics related to filtering. Not only information was the subject for filtering but also services opening a new field of service filtering and adding value to semantic business process management. Achieving the above mentioned goal required the development of techniques for information structuring as well as for representation of users needs. The latter was achieved by building profiles based on contents of information systems. The first was in spotlight of the project and produced a multitude of results. In both cases ontologies proved to deliver required level of formalisation. For structuring of information advanced information extraction techniques were used. The team worked with linguistic suite Sprout (originally developed by DFKI, Germany) and extended resources for Polish language: both gazetteer and grammar rules. The meaning of information depends on context; therefore, ENIRAF distinguished temporal, geographical and organisational contexts. Determination of context required development of respective sophisticated rules. ENIRAF focused mostly on Polish as it was neglected in the research until then. Proper structuring got in loop with ontology population from text. It opened a new challenge which was ontology evolution, closely related to the temporal context of information. The theoretical framework for ontology evolution, including formal notation, was proposed - contextual named triples. Additionally, a plug-in for Protégé was developed, which allows adding axioms to the ontology and removing them as necessary where axioms are generated from text documents. Besides the technical and foundational development, the project was also dealing with architectural issues, software design, and general computational limits that apply also for ontologies. One of the work packages was devoted to specific application area which is legal information, focusing particularly on insurance market and semantic representation of a risk. Key development of the project have been summarised in the book written by Prof. Witold Abramowicz (Filtrowanie informacji, Wydawnictwo Akademii Ekonomicznej w Poznaniu, 2008, 627 pp, ISBN 978-83-7417-315-5). The book 'Information filtering', written in Polish, summarises the research record of the author gained at Poznan university of Economics (Poland), Technical University of Zurich (Switzerland) and Humboldt university (Germany). It includes also results of numerous research projects the author participated in, with the strongest focus on ENIRAF. The book covers the results of interdisciplinary research in the fields of: advanced methods of information retrieval from the Internet, methods of knowledge representation, information needs formalisation of information systems users as well as filtering of information in the e-business and public administration domain. The entire book concludes with the vision of evolution of information filtering technology including different objects being filtered (e.g. text documents versus web services) as well as different surrogates (syntactic versus semantic representation) and in consequence different mechanisms that are used. The book was awarded a prize for the best academic book of 2008 in Poland. The project had also a strong impact on teaching and training, on all levels according to Bologne process. Most noticeable was doctoral level, where involvement of fellows in various forms resulted in raising of academic qualifications. Two doctoral dissertations were defended during the project and six other were well advanced so that the defences should take place soon after the end of the project. The book prepared by Prof. Abramowicz is considered a full professorship contribution.

Data: CORDIS, © European Union