Wikipedia is a public and universal encyclopedia where contributors edit articles collaboratively. Wikipedia infoboxes and categories have been used by semantic technologies to create DBpedia, a knowledge base that semantically describes Wikipedia content and makes it publicly available on the Web. Semantic descriptions of DBpedia can be exploited not only for data retrieval, but also for identifying missing navigational paths in Wikipedia. Existing approaches have demonstrated that missing navigational paths are useful for the Wikipedia community, but their injection has to respect the Wikipedia convention. In this paper, we present a collaborative recommender system approach named BlueFinder, to enhance Wikipedia content with DBpedia properties. BlueFinder implements a supervised learning algorithm to predict the Wikipedia conventions used to represent similar connected pairs of articles; these predictions are used to recommend the best convention(s) to connect disconnected articles. We report on an exhaustive evaluation that shows three remarkable elements: (1) The evidence of a relevant information gap between DBpedia and Wikipedia; (2) Behavior and accuracy of the BlueFinder algorithm; and (3) Differences in Wikipedia conventions according to the specificity of the involved articles. Blue-Finder assists Wikipedia contributors to add missing relations between articles, and consequently, it improves Wikipedia content.