PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

High-coverage extraction of semantic assertions from text
Mitja Trampus and Dunja Mladenić
In: IS 2011, 8-12 Oct 2011, Ljubljana, Slovenia.


We present an early version of a method for opendomain semantic assertion extraction from natural language texts. To combat the shortage of training data for the task, a two-stage pipeline is employed: we first perform semantic role labeling, then map the resulting frames onto predicate-form, ontology-aligned statements. We chose FrameNet and Cyc as the frame database and the ontology, respectively.

EPrint Type:Conference or Workshop Item (Paper)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Natural Language Processing
Information Retrieval & Textual Information Access
ID Code:8727
Deposited By:Jan Rupnik
Deposited On:21 February 2012