高相似度英语词语自主选取系统设计
2017-12-20李芳
李芳
摘 要: 针对传统的基于支持向量机的高相似度英语词语自主选取系统一直存在选取效果差、精度低的问题,提出一种基于数据挖掘的高相似度英语词语自主选取系统设计方法。首先根据英语词语的相似度概念,计算出两个英语词语义项的最短路径与其距离最近的公共父节点之间的深度。利用数据挖掘法将英语词语文本特征选择转换为一个多目标优化问题;然后以英语词语特征维数最少、分类正确率相对最高为选取标准,采用蚁群算法找到英语词语的最优特征子集;最后通过建立神经网络分类器完成高相似度英语词语自主选取系统设计。实验结果证明,所提方法可以精确地选取出高相似度英语词语,且选取时间较短,实用性广泛。
关键词: 高相似度英语词语; 数据挖掘; 自主选取; 系统设计
中图分类号: TN02?34; TP391 文献标识码: A 文章编号: 1004?373X(2017)23?0147?04
Abstract: Since the traditional high?similarity English words autonomous selection system based on support vector machine has the problems of poor selection effect and low accuracy, a design method of high?similarity English words autonomous selection system based on data mining is put forward. According to the similarity concept of English words, the shortest path of the semantic item between the two English words and its nearest depth among the common parent nodes are calculated. The data mining method is used to convert the text feature selection issue of English words into a multi?objective optimization problem. Taking the least English words feature dimensions and relatively?high classification accuracy as the selection standards, the ant colony algorithm is adopted to find out the optimal feature subset of English words. The neural network classifier is established to complete the design of the high?similarity English words autonomous selection system. The experimental results show that the proposed method can select the English words with high similarity accurately, and has short selection time and broad practicability.
Keywords: high?similarity English word; data mining; autonomous selection; system design
0 引 言
英语词语之间的语义相似度研究课题作为自然语言处理和计算机人工智能的基础性研究[1?3],如选取、搜索、分类以及歧义消除等,需要依赖于包含现实世界概念的知识体系[4]。英语词语的相似度是对英语词语之间语义相似紧密程度的度量,在机器翻译、数据信息检索等方面具有重要的实用价值[5]。在不同的实际应用场景中,英语词语相似度具有不同的用途[6]。另外,在构造统计英语词语语言模型的过程中,由于数据分布稀疏导致未登录英语词语的统计信息无法计算的问题[7],需要进行高相似度英语词语自主选取系统设计。……
