On "deep" knowledge extraction from documents

Udo Hahn, Martin Romacker
2000 Open research Areas in Information Retrieval  
S y n D i K A Te com prises a fam ily o f natural language understanding systems for automatically acquiring know l edge from real-w orld texts (e.g., information technology test reports, medical finding reports), and fo r transferring their content to formal representation structures which constitute a corresponding text know ledge base. We present a general system architecture w hich integrates requirem ents from the analysis o f single sentences, as w ell as those o f referentially linked
more » ... tences form ing cohesive texts. Properly accounting for text cohesion phenom ena is a prerequisite for the soundness and validity o f the generated text representation structures. It is also crucial for any inform ation system application m aking use o f automatically generated text know ledge bases in a reliable way.
dblp:conf/riao/HahnR00 fatcat:f7coedcddbabjngfs4iwzegv5i