One type of knowledge banks, named university where domain experts are driving forces in teaching, academic services and research. in big universities, there exist many domains such as agriculture, arts, economics, engineering, medical science, information technology, education, and social science are offered for studying. searching who is an expert in which area is lengthy. the objective of this study is to propose algorithm for clustering experts from their past studies and research work. data set were collected from employee profiles of a public university of thailand. series of data mining techniques including attribute selection, attribute weihting and two step data clustering by SOM, and k-means algorithm were used to segment experts into specific groups. these three candidate features weighting techniques include TF-IDF (term frequency-inverse document frequency), logarithm weight and augmented weight. it seemed that feature weighting could be used to improve clustering performances. recommendations for future research are provided.
|
|