Tagging-based systems are a popular and convenient way to organize information on the Web. Despite the alleged advantage of the free choice of words used to categorize Web resources in this kind of systems, it also brings some disadvantages due to the difficulty to remember freely chosen tags when users need to retrieve tagged resources. This paper presents a new approach to improve the quality of the categorizations performed by users in tagging-based systems by means of the recommendation of semantic tags. Our approach combines three sources of information for selecting the recommended tags: the Web resource been categorized, the tagging-based system folksonomy and the user personomy. By using these sources, we combine some features of the context of the categorization, the social opinion about the resource been categorized and the users’ vocabulary preferences. The use of the Web resource helps to solve the cold start problem, and the recommendation of more contextualized and personalized tags helps to develop a better personomy for the user, which could relieve the users’ cognitive effort when retrieving tagged resources.