In today's world, the volume of literature is growing at an almost exponential rate, yet documents are rarely distributed purely through electronic data transfer.

Due to various constraints, a document sometimes travels in a spiral: Word -> PDF -> a printout on paper -> scanned back into a file -> and only then reorganized. And when it comes time to reorganize the material, one inevitably encounters all kinds of strange data—1 turning into i, some fields completely misaligned, and so on.

ParsCit, an open-source package, attempts to take the references at the end of a document and restore them into well-ordered data using CRF-based analysis.

This paper describes the underlying principles. The official site is: http://wing.comp.nus.edu.sg/parsCit/ The latest download is currently at: http://wing.comp.nus.edu.sg/parsCit/parscit-110505b.zip

$ wget -c  http://wing.comp.nus.edu.sg/parsCit/parscit-110505b.zip \
 -o wget-log -r --tries=50 -v -O parscit-110505b.zip

Since it serves as a reference, parts of my translation draw on this.