An Approach to Modelling User Interests using TF-IDF and Fuzzy Sets Qualitative Comparative Analysis Dimitris K. Kardaras 2 [ORCID] and Stavros Kaperonis 1 [0000-0002-2130-6514] and Stavroula Barbounaki 3[ORCID] and Ilias Petrounias 4[ORCID] and Kostas Bithas 5[ORCID] 1 Athens University of Economics and Business, Patission str. 29, 10434 Athens, Hellas dkkardaras@yahoo.co.uk 2 Panteion University of Social and Political Sciences, Syggrou ave. 136, 17671 Athens, Hellas skap@panteion.gr 3 Merchant Marine Academy of Aspropyrgos, Dragatsaniou str. 8, 10559, Athens, Hellas sbarbounaki@yahoo.gr 4 The University of Manchester, Oxford Rd M13 9PL, Manchester, UK ilias.petrounias@manchester.ac.uk 5 Panteion University of Social and Political Sciences, Syggrou ave. 136, 17671 Athens, Hellas kbithas@eesd.gr Abstract. Modelling and understanding user interests are particularly important tasks for designing services and building systems for customized solutions in web personalization and recommender systems. User generated content (UGC) constitutes a significant source of information for capturing user interests. This paper, suggests an approach to user profiling that analyses the TF and IDF factors of selected tourism services by utilising the Fuzzy set Qualitative Comparative Analysis (FsQCA). It analyses a sample of customer reviews that are collected from tourism web sites. This paper considers the amount of money that customers spent during their hotel stay, as the outcome set in the FsQCA analysis. The results produce causal combinations of services that are necessary and sufficient for building customer interests models that best lead to the outcome and argue for the applicability of the FsQCA in modelling user interests. Keywords: User interests, Fuzzy Sets Qualitative Comparative Analysis, TF, IDF 1 Introduction A recommender system (RC) provides users with personalized suggestions for online products or services to improve customer relationship management. It is well grounded that RC represent a great opportunity for developing customised services across business sectors such as government, tourism, education, media [1]. RC utilise techniques spreading from statistics, to AI and machine learning in order to capture user interests, build user and products/services profiles and suggest the most appropriate products or services to them. RC do not only rely on demographics and usage data but also utilise knowledge so that they take into consideration issues pertaining to user’s mood and phycology [2]. RC contribute with information filtering by providing the users with the most relevant information and services. RC draw on several methods that can assist in developing user references models, with user-generated-content (UGC) to represent a source with rich customer information [3]. As such, social media platforms help users to exchange experience, feedbacks, opinions, complaints, thus providing significant information for capturing and understanding user interests [3]. The importance of UGC analysis is also highlighted by [4], arguing that user reviews can provide new ways of understanding user needs, establishing communication channels with customers and managing companies’ online reputation. Web personalisation is another area where user profiling is necessary for developing customised web interfaces, supporting personalised search [5]. Personalised search refers to ability of retrieving and filtering search results according to individual user’s personal needs [5]. In a number of cases, recommendation systems also take on the dynamic context of the trip in addition to the traveller’s preferences [6]. 2 User Profiling in Tourism Building user interests models has also been the focus of e-tourism research studies. Drawing on behavioural, socio-economic and demographic data analysis several researchers shed light into understanding people's travel