Existing semantics-based word sense disambiguation method has two shortcomings: First, it is unreasonable to use some of the contextual words with ambiguity for disambiguation. Second, it does not consider the effect of contextual words' distance on semantic relatedness calculation, the article puts forward an improved method in view of this, which uses two approaches, step-by-step strategy and distance weighting. Experimental results show that the improved method has remarkable amelioration in disambiguation effect.%针对现有基于语义的词义消歧方法存在两点不足:一,利用部分具有歧义的上下文语境词进行消歧存在不合理性;二,未考虑上下文语境词距离远近对语义相关度计算的影响,提出一种改进的方法,采用分步策略和距离加权两种方法分别进行改进.实验结果表明,改进方法在消歧效果上有明显的改善.
展开▼