Semantic analysis of documents workshop (SemADoc): extended abstract

Abstract

A large number of document management problems would benefit from having the semantics of documents explicitly represented. However, manually assigning semantic descriptions to documents is labour intensive and error prone. At the same time, the manual generation of domain specific taxonomies is not only labour intensive, but it also needs to be repeated often as the domains themselves and their key concepts shift with time. In this workshop we focus on document content analysis and semantic enrichment to generate a layer of semantic description of documents that is useful for document management tasks, such as semantic information retrieval, conceptual organization and clustering of document collections for sense making, semantic expert profiling, and document recommender systems. The aim of the workshop is to bring together researchers and practitioners, and discuss different perspectives on the problems, challenges encountered in various application scenarios, and potential solutions. We have invited submissions in all areas of semantic analysis and enrichment of documents, such as automatic tagging, named entity disambiguation, semantic linking, interactive classification and clustering of documents, document summarization, curation and validation of the analysis process, generation of visualizations of document, author and document collection semantics, user engagement in the semantic analysis process via suitable annotation and correction tools, and study of the trade off between accuracy of the results and user effort. Submissions aimed at solving practical problems in specific application domains, including but not limited to digital libraries, legal document management, personalized online learning systems, news media, are especially welcome. The workshop is timely and relevant to the Document Engineering community, as its focus is on semantically enriching documents and document collections, to make them more accessible to their readers. The task is nontrivial due to the volume of text data and the rate at which text data is accumulated by companies, government, and individuals.