Fast Data Mining For Programming Support – LWA 2004

by Thomas Zimmermann

We apply association rule mining to version histories: "Programmers who changed these functions also changed..." Such rules suggest and predict likely further changes and prevent errors due to incomplete changes. To deal with the required low support values we use constraints on the rules and mine on the fly. The topmost three suggestions of our ROSE prototype contain a correct location with a likelihood of 64%.

See also: http://www.softevo.org/

Reference

Thomas Zimmermann. Fast Data Mining For Programming Support. In LWA 2004: Lernen - Wissensentdeckung - Adaptivität, Berlin, 4. - 6. Oktober 2004, Workshopwoche der GI-Fachgruppen/Arbeitskreise (1) Fachgruppe Adaptivität und Benutzermodellierung in Interaktiven Softwaresystemen (ABIS 2004), (2) Arbeitskreis Knowledge Discovery (AKKD 2004), (3) Fachgruppe Maschinelles Lernen (FGML 2004), (4) Fachgruppe Wissens- und Erfahrungsmanagement (FGWM 2004) (LWA 2004), Berlin, Germany, October 2004, pp. 100-109.

BibTeX Entry

@inproceedings{zimmermann-lwa-2004,
    title = "Fast Data Mining For Programming Support",
    author = "Thomas Zimmermann",
    year = "2004",
    month = "October",
    booktitle = "LWA 2004: Lernen - Wissensentdeckung - Adaptivität, Berlin, 4. - 6. Oktober 2004, Workshopwoche
der GI-Fachgruppen/Arbeitskreise (1) Fachgruppe Adaptivität und Benutzermodellierung in Interaktiven
Softwaresystemen (ABIS 2004), (2) Arbeitskreis Knowledge Discovery (AKKD 2004), (3) Fachgruppe Maschinelles Lernen (FGML
2004), (4) Fachgruppe Wissens- und Erfahrungsmanagement (FGWM 2004)",
    location = "Berlin, Germany",
    pages = "100--109",
    publisher = "Humbold-Universität Berlin",
}