In this paper, a novel machine learningapproach for the identification of namedentity relations (NERs) called positiveand negative case-based learning(PNCBL) is proposed. It pursues the improvementof the identification performancefor NERs through simultaneouslylearning two opposite cases and automaticallyselecting effective multi-levellinguistic features for NERs and non-NERs. This approach has been applied tothe identification of domain-specific andcross-sentence NERs for Chinese texts.The experimental results have shown thatthe overall average recall, precision, andF-measure for 14 NERs are 78.50%,63.92% and 70.46% respectively. In addition,the above F-measure has been enhancedfrom 63.61% to 70.46% due toadoption of both positive and negativecases.
展开▼