We used automatic text-mining of PubMed abstracts of papers related to obesity, with the aim of revealing that the information used in abstracts reflects the current understanding and key concepts of this widely explored problem. We compared expert data from DisGeNET to the results of an automated MeSH (Medical Subject Heading) search, which was performed by the ScanBious web tool. The analysis provided an overview of the obesity field, highlighting major trends such as physiological conditions, age, and diet, as well as key well-studied genes, such as adiponectin and its receptor. By intersecting the DisGeNET knowledge with the ScanBious results, we deciphered four clusters of obesity-related genes. An initial set of 100+ thousand abstracts and 622 genes was reduced to 19 genes, distributed among just a few groups: heredity, inflammation, intercellular signaling, and cancer. Rapid profiling of articles could drive personalized medicine: if the disease signs of a particular person were superimposed on a general network, then it would be possible to understand which are non-specific (observed in cohorts and, therefore, most likely have known treatment solutions) and which are less investigated, and probably represent a personalized case.