基于本体的通信服务网络抱怨案例相似度计算

现有的基于本体的抱怨处理方法大多都依赖企业员工概括当前的顾客抱怨为关键词并用于检索解决方案,故不适用于数量巨大且传播速度快的网络抱怨.针对这个问题,本文引入相似性度量技术,提出一种基于本体的通信服务行业网络抱怨案例相似度计算方法.构建该行业的网络抱怨案例本体,基于该本体,建立聚合抱怨内容相似度、抱怨产品相似度和抱怨问题相似度的抱怨案例相似度模型;在理论上证明所建立的聚合模型的高准确性,并在实验中验证其高准确性,同时还使用平均绝对误差、准确率、精确度、召回率和F1值对其性能进行评价.理论证明结果表明,聚合模型比基准模型准确性更高.实验验证结果表明,聚合模型比基准模型准确性更高且性能更佳.

Ontology-based similarity measurement for online complaint cases of communications services

Most of existing ontology-based complaint handling methods rely on employees to summarize some keywords from a current customer complaint and to use these keywords to retrieve for a corresponding solution, but these methods cannot be suitable for online complaint, which has large quantity and fast velocity of propagation. Aiming at this problem, this paper introduces similarity measure technology into the existing ontology-based online complaint handling method, and proposes an ontology-based similarity measure method for online complaint cases of communications services industry. An online complaint case ontology for communication services industry is constructed. Based on this ontology, an aggregation model of similarity between complaint cases is established through combining the similarities of complaint contents, complaint products and complaint problems. The high accuracy of the established aggregation model is theoretically proved and experimentally verified. Further, the indicators like mean absolute error, accuracy rate, precision, recall and F1 are used in the experiment to evaluate the performance of the aggregation model. The proof result suggests that the accuracy of the aggregation model is higher than the accuracies of baseline models. The experiment result shows that the same conclusion and also shows that the performance of the aggregation model is better than the performances of baseline models.