The purpose of this paper is to describe the scope, approach, methods and algorithms for the development and implementation of a data mining system using the Symptom Identification, Modeling and Context Observation (SIMCO) technique, designed to analyze cognitive and decision-making errors during the regular training sessions of the Main Control Room (MCR) crews of the full-scale simulator (FSS-1000) for a pressurized water reactor (PWR), type VVER-1000. The pilot implementation of the SIMCO system at the Kozloduy nuclear power plant (NPP), through experimental testing, validation and verification of the methodology, will also enable the automated data mining of offline and online assessment of the symptom-based emergency operating procedures (EOP) and crews' practical skills. The scope of the methodology is not limited to the activities of collecting data for human reliability assessment (HRA), processing and analyzing these data to assess the human error probability (HEP) in a mode and format for their integration in the probabilistic safety analysis (PSA) of units 5 and 6 of the Kozloduy NPP. This methodology can also complement the quality assessment of human factors engineering and the effectiveness of the human-machine interface under normal and abnormal operating conditions, as well as operators’ theoretical knowledge and abilities, making the personnel testing and licensing process at the Bulgarian Nuclear Regulatory Agency (BNRA) more systematic and objective.