This paper proposes ``a data rearrangement engine'' for data mining to propose a framework for realizing efficient data manipulation in data mining. The difference between a data rearrangement engine and a relational database engine is that the primary focus of the former is on the data manipulation called data rearrangement. In this paper, we propose four basic operators in for the data rearrangement engine, namely grouping, attribute addition, rearrangement, and fetch, and introduce how they can be used for describing various queries. Moreover, we propose a query language for accessing the data rearrangement engine, and illustrate their effective and flexible usages for image retrieval, clustering, and time series data.

      @InProceedings{ k:prmu03-2,
      author = {Asanobu KITAMOTO},
      title = {A Data Rearrangement Engine for Data Mining},
      booktitle = {Technical Report of Institute of Electronics, Information, and Communication Engineers (IEICE)},
      volume = {PRMU2003-78},
      pages = {19-24},
      year = 2003,
      month = 9,
      doi = {},
      note = { (in Japanese)},
      }