Researches focused on human emotions identification at work with computer reach increasing care in the area of human computer interaction and artificial intelligence. Researchers aim to recognize user´s emotions the most exactly. Condition of solution is that emotions influence work and behavior of person during the time of using computer. In this paper we are focused on user´s model biometric characteristics and its influence determination for recognition of emotions at work with computer. We

more »

... ropose method for influence determination of each particular characteristic inspired by calculating information gain used in classification. On the base of influence and according to properties of these characteristics we have defined own weighting coefficient, which we apply in comparative methods for emotion´s recognition. We show influence of user characteristics and how our approach increases success of emotion recognition. INTRODUCTION In this paper we are focused on the user´s model for the aim of his emotions and emotional status identification. Mood, emotions, feelings and states of human are changeable in various situations, also during the time of using computer. Occur of negative emotion at work with computer is highly frequent. This fact was confirmed by symposium that we realized. Our respondents said that they have felt states as tiredness and stress periodically. But computer systems are not able to understand and adapt to user. Because of this reason computer programs react unsuitable, provide incorrect feedback, applications interrupt work and it lead to user increasing frustration. So, after successful recognition of actual user emotion in computer environment, solutions want to offer better and more effective conditions of his work. Our general goal is recognition of emotions by modeling user according his biometric characteristics. In solution we come out from fact, that emotions influence human work and behavior during using computer. User behavior and manners affect biometric characteristics such as keystroke dynamics, mouse dynamics and work with applications. These characteristics we can monitoring and then create model from characteristics values. We understand user model as mathematical formula of person behavior that allow us to recognize user or user emotions. Identification is realized by comparing created model with data captured from real time work with computer. In this paper we are mainly focused on selection of the most relevant user characteristics and their weighting, that lead to more exact emotion recognition.