I'm looking to build an undirected network of chemicals extracted from biomedicine papers, specifically from MEDLINE. In this network, the nodes would be chemicals and the links would determine publications which mentioned two or more chemicals together. Eg., if A and B are two chemicals, then a link between them would correspond to a paper that mentioned them together. More the papers featuring those two chemicals, greater the link weight between the two corresponding nodes. A paper mentioning 5 chemicals would form a total of 15 links between the 5 nodes - connecting a pair at once. There are tools to extract words, citations, authors, etc. from journal metadata for analysis (like VOSViewer). Is there a way that I can extract the chemical names from titles and abstracts of MEDLINE papers?