作者: Ricardo Perez-Castillo , Laura S´nchez-Gonz´lez , Mario Piattini , Felix Garcia , Ignacio Garcia-Rodriguez de Guzm´n
DOI: 10.1109/ESEM.2011.64
关键词:
摘要: Business process mining is a powerful tool to retrieve the valuable business knowledge embedded in existing information systems. The effectiveness of this kind proposal usually evaluated using recall and precision, which respectively measure completeness exactness retrieved processes. Since assessment difficult error-prone activity, main hypothesis work studies possibility obtaining thresholds determine when precision values are appropriate. technique under study MARBLE, model-driven framework processes from Bender method was applied obtain measures. experimental data used as input were obtained set 44 with MARBLE through family case carried out over last two years. provides for measures, facilitates interpretation their by means five linguistic labels that range low very high. As result, must be high (with at least medium above 0.56), also 0.70) ensure recovered (by MARBLE) an value 0.65. allowed us ascertain more confidence whether can effectively mine In addition, provided results reference compare other similar techniques.