1. We would like a robust way to verify all citations and links of a paper, and flag out fake / invalid ones 2. This will involve parsing user's project structure and various preprocessing to optimize identification