@misc{DomenachFrankeMaierChristofidisetal.2016, author = {Domenach, Florent and Franke-Maier, Michael and Christofidis, Christos and Wartena, Christian}, title = {Dynamic semantic subject indexing completion for classification of library items}, url = {http://nbn-resolving.de/urn:nbn:de:0290-opus4-24475}, year = {2016}, abstract = {Subject indexing is an essential part of categorizing library items, but can be unreliable, inconsistent or missing, especially in ebook collections bought as a whole. The result is a lack of relevant searchable knowledge units for the use of library patrons. Completing those subjects manually is time consuming, especially when large corpora are considered. Formal Concept Analysis (FCA), which builds a concept lattice from a binary context, can be used to support such a knowledge discovery task. Pattern structures enable a general FCA-based framework to build a concept lattice from complex data, e.g. the subject terms of a library resource. Our approach, based on this general framework, dynamically completes a book´s subject terms in order to improve data integration. If subject terms are altogether missing, we use WordNet as reference ontology to extract keywords from the title words of the library item and map them to the terms of the library authority files. References: A. Coulet, F. Domenach, M. Kaytoue and A. Napoli (2013), Using pattern structures for analyzing ontology-based annotations, Formal Concept Analysis, P. Cellier, F. Distel and B. Ganter (eds), LNCS/LNAI vol. 7880, pp 76-91 Ganter, B., Kuznetsov, S. O.: Pattern Structures and Their Projections. ICCS, Springer, 2001, 2120, 129-142 Ganter, B., Wille, R.: Formal Concept Analysis, Mathematical Foundations, Springer Verlag (1999)}, language = {de} }