Grammatical Case Based IS-A Relation Extraction with Boosting for Polish
This work addresses the challenge of extracting taxonomical relations for Polish language processing, representing an incremental improvement over existing pattern-based methods.
The paper tackles the problem of extracting IS-A relations from text by introducing a novel method that uses morpho-syntactical annotations and grammatical case of noun phrases, along with pseudo-subclass boosting to increase extraction. Experiments on a corpus of 0.5 billion Polish web documents showed improved relation extraction, though no specific numbers for performance gains are provided.
Pattern-based methods of IS-A relation extraction rely heavily on so called Hearst patterns. These are ways of expressing instance enumerations of a class in natural language. While these lexico-syntactic patterns prove quite useful, they may not capture all taxonomical relations expressed in text. Therefore in this paper we describe a novel method of IS-A relation extraction from patterns, which uses morpho-syntactical annotations along with grammatical case of noun phrases that constitute entities participating in IS-A relation. We also describe a method for increasing the number of extracted relations that we call pseudo-subclass boosting which has potential application in any pattern-based relation extraction method. Experiments were conducted on a corpus of about 0.5 billion web documents in Polish language.