A Study of Data Clustering Method Cooperate with Expert Knowledge

• Main Authors: Yao Hui-Lin, 姚輝林
• Other Authors: Wang Wei-Hua Andrew
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/59573961926055167594

碩士 === 東海大學 === 工業工程與經營資訊學系 === 92 === Clustering has its root in many areas, including data mining, statistics, biology, and machine learning. Clustering is the process of grouping data into classes. The objects within a cluster have high similarity in specific features comparison with each another, yet, are very dissimilar to objects in other clusters. In the past, lots of the clustering algorithms are proposed. Dissimilarities are assessed based on the attribute values describing the objects. In general, the differences are characterized by similarity measures. However, it used the properties of data to be the basis of analysis in most clustering algorithms. Since data can only provide the relationships between attributes, but not the causalities. The lack of domain knowledge causes to some clustering mistakes. In this paper, we provide the mechanism to cooperate the data clustering algorithm with the expert''s domain knowledge. This research proposed a new similarity measure; the measure can set each attribute with different weight to calculate the similarity. This research adds the expert''s domain knowledge into the data clustering algorithm by means of a new similarity measure. The ZOO dataset and the PET dataset are used to validate the effectiveness of the new algorithm.