This research was done as our term project for EE 361M - Introduction to Data Mining taught by Dr. Joydeep Ghosh.
Automating the assignment of MeSH terms to biomedical articles is a topic that has been heavily researched, due to the high cost of manually assigning terms. Here, we have taken a novel approach at this task to facilitate automatic indexing using the cited works of a document. Similar to some recent work, we approach the problem as a ranking problem so that MeSH terms more relevant to the target document are ranked higher. Specifically, our method is to use the MeSH terms from both the nearest neighbors and the cited works of a target article and rank these terms by relevance.
For more information on this research topic, see here