Electronic negations are negotiations conducted through electronic means. In electronic negotiations text messages are the main tool used by the participants in achieving their goals. The texts convey influence strategies, e. g., persuasion, and role-dependent behaviour, e.g., buyer's behaviour.
In this talk we focus on modelling and classification of the negotiation texts. We discuss the results of statistical language models. Selection of text features for Machine Learning is discussed after that. We show correlation between the tasks of learning, Machine Learning tools and selected data features. We also address the problem of prediction (learning on the earlier stages of negotiations).