论文信息 - PolyAnalyst Data Analysis Technique and Its Specialization for Processing Data Organized as a Set of Attribute Values

PolyAnalyst Data Analysis Technique and Its Specialization for Processing Data Organized as a Set of Attribute Values

Abtract. The data analysis techniques of the PolyAnalyst data mining system [2] are based on the automated synthesis of functional programs treated as the multi-dimensional non-linear regression models. This approach provides the system with two valuable properties: 1) it can discover in data the hidden relations that might be of a great variety of forms, 2) it can explore arbitrarily complexly structured data if the corresponding data access primitives are provided. The paper contains a formal description of the final version of the basic PolyAnalyst mechanisms, which are utilized in the general case, as well as in a particular case of data organized as a set of attribute values (SAV), which is the most common format for data explored by KDD methods.

[1] Mikhail V. Kiselev,et al. Regression-Based Classification Methods and Their Comparisons with Desision Tree Algorithms , 1997, PKDD.

[2] John W. Backus,et al. Can programming be liberated from the von Neumann style?: a functional style and its algebra of programs , 1978, CACM.

[3] Jieming Zhu,et al. Panel: Application of Empirical Discovery in Knowledge Acquisition , 1991, EWSL.

[4] Toby Walsh,et al. Proceedings of AAAI-96 , 1996 .

[5] Jan M. Zytkow,et al. Discovery of Equations: Experimental Evaluation of Convergence , 1992, AAAI.

[6] BackusJohn. Can programming be liberated from the von Neumann style , 1978 .

[7] Wei-Min Shen,et al. Functional transformations in AI discovery systems , 1988, [1988] Proceedings of the Twenty-First Annual Hawaii International Conference on System Sciences. Volume III: Decision Support and Knowledge Based Systems Track.

[8] Mikhail V. Kiselev,et al. PolyAnalyst - A Machine Discovery System Inferring Functional Programs , 1994, KDD Workshop.