<p>Researchers and practitioners from different disciplines (e.g., educational science, computer science, statistics) continuously enter the rapidly developing research field of learning analytics (LA) and bring along different perspectives and experiences in research design and methodology. Scientific communities share common problems, concepts, theories, world views, and methodologies. The goals of this research are to identify scientific communities among the researchers working on the LA use cases in higher education, as reflected in their common publications, to detect common themes in these publications, and to track the development of themes and communities in the first ten years since the first Learning Analytics Conference. A systematic literature review, followed by a combination of natural language processing for the analysis of themes, and social network analysis of co-authorship networks for the detection of scientific communities were used. Conducted analyses revealed 10 most common themes in research field of LA where six out of 10 themes appeared in more than 20 papers. Themes that appeared the earliest are those that are related to the use of machine learning for prediction of academic performance and those related to students’ learning behavior. The most recent themes are related to feedback and assessment, as well as LA adoption. Also, two large scientific communities and the most prominent researchers among them are detected. The integrating mechanisms of these scientific communities were mapped to the theories of market of scientific credibility and political technologies. Measures of network centrality of authors correspond well with the usual bibliometric indicators of research impact.</p>