移动网络相似信息重复记录智能检测仿真

Simulation of Duplicate Records Intelligent Detection of Similar Information in Mobile Network

移动网络相似信息重复记录检测在专利分析系统中具有广泛的应用前景。针对当前方法存在检测耗时较长、查准率和查全率较低等问题,提出一种基于领域本体的移动网络相似信息重复记录智能检测方法,构建了一种三维的移动网络文本空间表示模型,对移动网络中相似信息重复记录文本集合中的文本向量进行结构化描述。在此基础上,基于领域本体分别对移动网络相似信息重复记录中的词语、句子和文本进行相似度检测,得到移动网络文本中任意两个句子的相似度特征矩阵。对移动网络文本中句子相似度特征矩阵进行遍历,选取其中相似度最大的句子组合,并将该组合所属行列从矩阵中删除,再从剩余矩阵中相似度最大的句子组合筛选出来,以此类推,直到句子中的元素数目变为0,提取获得相似度最大句子组合序列,根据该序列即可实现移动网络相似信息重复记录的智能检测。仿真测试结果表明,上述方法在移动网络相似信息重复记录相似度检测准确性上更具优势,具有较高的查准率和查全率,并且检测效率较高。

Due to the time consuming detection and the low precision and recall rate in current methods, this paper puts forward the intelligent detection method of similar information in mobile network based on domain ontology. At first, this method constructed a three-dimensional representation model of text space in mobile network. In the model, the text vectors in the text set of duplicate record of similar information were structurally described. On this basis, similarity detection was performed on words, sentences and texts in duplicate record of similar information of mobile network. Then, the similarity feature matrix of any two sentences in the mobile network text was obtained. Moreover, similarity feature matrices of sentence in mobile texts were traversed to choose the sentence combination with the maximum similarity, and the row and the column which the combination belonged to were deleted from the matrix, and then sentence combinations with the maximum similarity were selected from remaining matrices until the number of elements in the sentence became zero. Finally, the sequence of sentence combination with maximum similarity was extracted. According to this sequence, the intelligent detection of duplicate record of similar information in the mobile network was realized. Simulation results show that the above method has some advantages on the accuracy of similarity detection for duplicate record of similar information in mobile network. Meanwhile, it has higher precision rate and recall rate, and the detection efficiency is high.