How matching works
Your text is compared with publication titles, eligible CC-BY/CC0 abstracts and supervised-work titles using TF-IDF and cosine similarity. Shared words and phrases drive the ranking. Independent evidence is combined, with supervised-work evidence weighted at 0.8.
Names are removed before matching. The top five results show their supporting sources. Similarity values are relative scores, not probabilities or a measure of supervision quality.
Read the evidence, then reach out
Faculty coverage varies by school. A missing publication record does not imply a lack of expertise. Some rosters include researchers, invited or retired faculty; check current roles, eligibility and capacity with the school.
Contact drafts use your proposal and the listed evidence. They are editable and open in your email app; nothing is sent automatically. Matching runs on your device. Your research text is never sent to our server or saved by this site. The first search downloads a reusable index; later searches use the same local index. It is an independent project, not an official NOVA service.