Towards an Automated Retrieval of Publications based on Telescope Observations

We analyze papers based on VLT observations regarding (a) the information provided about facilities used and (b) possible retrieval through ADS. The following topics are looked at:

* Where in the papers do authors report on the facilities used and how detailed is this information? Can it be used to measure scientific output of observatories, telescopes, instruments and observing programs?

* Which percentage of relevant papers is retrieved through ADS, how many irrelevant papers are picked up, and why?

* How can automated retrieval of papers be improved and which factors remain problematic?

We conclude that the currently available retrieval options are not sufficiently reliable to abandon manual literature screening. The implementation of dedicated LaTeX-tags in the macros of core astronomy journals and a specific search option through ADS would probably improve the procedure considerably and help to move towards (mainly or entirely) automated retrieval of papers.