The purpose of this paper is to illuminate the perspective that tourists and particularly Greeks demonstrate as far as their booking behavior concerned travel reviews during the period of economic recession. Especially online travel reviews written by consumers are ever more available and used to inform travel-related decisions. The present research concentrated on the web platforms of intermediary firms and agencies that constitute the e-tourism sector. These are organizations that facilitate communication and transactions between the primary providers of travel and accommodation services (e.g. airlines, hotels, car hire firms), and potential consumers of those services. Blogs, online reviews (ORs), and social networking platforms are enabling travelers to share information, opinions, and knowledge about all kinds of goods and services in e-Tourism. Specifically, online reviews (ORs) can be considered as electronic versions of traditional WOM (e-WOM) and consist of comments published by travelers. For the data collection, three self- report questionnaires were administered: a) E-WOM and Accommodation Scale (E-WOM), b) Emotion-Based Decision-Making Scale (EBDMS,) c) Emotional Intelligence Questionnaire (TEIQue) clarifying four factors of emotional state; well-being, self-control, emotionality and sociability. There were being created their electronic versions through Google Forms service and posted through the website “http://www.cicos.gr/iccmi2017/obsr”. Then the collected data were selected for analysis, with relevant transformations in order to have a suitable form for the implementation of the respective machine learning algorithms included in the software package R. The methodology, that was adopted, consists of two concrete phases. During the first phase, questionnaires were created and submitted. During the second phase, the data set were collected, preprocessed and analyzed based on Data Mining techniques evaluating the results. More specifically, classification algorithms were utilized so as to manage to describe hidden patterns. Also, the parameters of the algorithms were set, depending on the application cases, and the results were correlated with the demographic characteristics of the respondents, in order to evaluate and assess the significance of exported rules / conclusions
