Automatic detection of causal relations for Question Answering

Abstract

Causation relations are a pervasive feature of human language. Despite this, the automatic acquisition of causal information in text has proved to be a difficult task in NLP. This paper provides a method for the automatic detection and extraction of causal relations. We also present an inductive learning approach to the automatic discovery of lexical and semantic constraints necessary in the disambiguation of causal relations that are then used in question answering. We devised a classification of causal questions and tested the procedure on a QA system.

Extracting Explicit and Implicit Causal Relations from Sparse, Domain-Specific Texts

“…Our scores of precision (76.5%) and recall (82%) compare favorably with those reported by other state-of-the-art algorithms [2,4,7,8,9,14]. These latter techniques, however, extensively relied on manually-crafted knowledge (e.g.…”

“…In addition, these texts are rife with implicit causal relations, which are realized by implicit verbal and non-verbal causal patterns. Implicit patterns and relations are more complex and difficult to detect than their explicit counterparts [7], traditionally extracted by current algorithms. Furthermore, extant algorithms are unable to precisely disambiguate ambiguous causal relations, which are realized by polysemous patterns.…”

“…Their corresponding recall scores are respectively 68% and 75.9%. In the supervised approach of [7], sentences in the L.A. Times corpus that contain explicit causal verbs (e.g. "to cause") are manually annotated as positive or negative examples of causal relations.…”

“…Causal relations are fundamental in many disciplines, including philosophy, psychology and linguistics. In Natural Language Processing (NLP), algorithms have been developed for discovering causal relations from large generalpurpose [2,4,7,8,14] and bio-medical corpora [9]. These algorithms rely extensively on hand-coded knowledge (e.g.…”

“…annotated corpora), and only extract explicit causal relations. Explicit relations are realized by explicit causal patterns, predominantly assumed to be causal verbs [4,7,14]. Causal verbs (e.g.…”