國立臺灣大學電機工程學系Li, Chung-ShengChung-ShengLiChang, Yuan-ChiYuan-ChiChangBergman, Lawrence D.Lawrence D.BergmanSmith, John R.John R.Smith2006-09-272018-07-062006-09-272018-07-062000http://ntur.lib.ntu.edu.tw//handle/246246/20060927122753664431In this paper, we describe a new paradigm for information retrieval in which the retrieval target is based on a model. Three types of models – linear, finite state, and knowledge models are discussed. These information retrieval scenarios often arise from applications such as environmental epidemiology, oil/gas production and exploration, and precision agriculture/forestry. Traditional model-based data and information processing usually requires the processing of each and every data points. The proposed new framework, in contrast, will process the data progressively using a set of progressive models and utilize indexing techniques specialized for the model to facilitate retrieval, thus achieving a dramatic speedup.application/pdf82279 bytesapplication/pdfzh-TWthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/20060927122753664431/1/0121.pdf