Schulz Stefan, Schober Daniel, Tudose Ilinca, Stenzhorn Holger
Medical Informatics Group, University Medical Center Freiburg, Germany.
AMIA Annu Symp Proc. 2010 Nov 13;2010:727-31.
Thesauri that are "ontologized" into OWL-DL semantics are highly amenable to modeling errors resulting from falsely interpreting existential restrictions. We investigated the OWL-DL representation of the NCI Thesaurus (NCIT) in order to assess the correctness of existential restrictions. A random sample of 354 axioms using the someValuesFrom operator was taken. According to a rating performed by two domain experts, roughly half of these examples, and in consequence more than 76,000 axioms in the OWL-DL version, make incorrect assertions if interpreted according to description logics semantics. These axioms therefore constitute a huge source for unintended models, rendering most logic-based reasoning unreliable. After identifying typical error patterns we discuss some possible improvements. Our recommendation is to either amend the problematic axioms in the OWL-DL formalization or to consider some less strict representational format.
被“本体化为”OWL-DL语义的叙词表极易出现因错误解释存在限制而导致的建模错误。我们研究了美国国立癌症研究所叙词表(NCIT)的OWL-DL表示形式,以评估存在限制的正确性。我们抽取了354条使用someValuesFrom运算符的公理作为随机样本。根据两位领域专家进行的评级,如果按照描述逻辑语义进行解释,这些示例中大约一半,进而OWL-DL版本中超过76,000条公理,会做出错误断言。因此,这些公理构成了意外模型的巨大来源,使得大多数基于逻辑的推理不可靠。在识别出典型错误模式后,我们讨论了一些可能的改进措施。我们的建议是要么修正OWL-DL形式化中的问题公理,要么考虑一些不太严格的表示格式。