Given the users from a social network site, who have been tagged with a set of terms, how can we recommend the movies tagged with a completely different set of terms hosted by another website? Given the users from a website dedicated to Type I and Type II diabetes, how can we recommend the discussion threads from another website dedicated to gestational diabetes, where the keywords used in the two websites might be quite diverse? In other words, how can we recommend across heterogeneous domains characterized by barely overlapping feature sets? Despite the vast amount of existing work devoted to recommendation within homogeneous domains (e.g., with the same set of features), or collaborative filtering, emerging applications call for new techniques to address the problem of recommendation across heterogeneous domains, such as recommending movies hosted by one website to users from another website with barely overlapping tags. To this end, in this paper, we propose a graph-based approach for recommendation across heterogeneous domains. Specifically, for each domain, we use a bipartite graph to represent the relationships between its entities and features. Furthermore, to bridge the gap among multiple heterogeneous domains with barely overlapping sets of features, we propose to infer their semantic relatedness through concept-based interpretation distilled from online encyclopedias, e.g., Wikipedia and Baike. Finally, we propose an efficient propagation algorithm to obtain the similarity between entities from heterogeneous domains. Experimental results on both Weibo-Douban data set and Diabetes data set demonstrate the effectiveness and efficiency of our algorithm.