Abstract. The core information from scientific publications is encoded in natural language text and monolithic documents; therefore it is not well integrated with other structured and unstructured data resources. The text format requires additional processing to semantically interlink the publications and to finally reach interoperability of contained data. Data infrastructures such as the Linked Open Data initiative based on the Resource Description Framework support the connectivity of data from scientific publications once the identification of concepts and relations has been achieved, and the content has been interconnected semantically. In this manuscript we produce and analyze the semantic annotations in scientific articles to investigate on the interconnectivity across the articles. In our initial experiment based on articles from PubMed Central we demonstrate the means and the results leading to the interconnectivity using annotations of Medical Subject Headings concepts, Unified Medical Language System terms, and semantic abstractions of relations. We conclude that the different methods would contribute to different types of relatedness between articles that could be later used in recommendation systems based on semantic links across a network of scientific publications.