Hello, We at PAHMA have developed a tool that I'd like to let the community know about, in case other institutions out there would like to use it or would like to work with us to improve it. The tool in question produces nightly emailed reports that detail the changes that have been made to CSpace data over the past day. We've had a couple of situations over the past year where we learned a bit belatedly that some users were making incorrect assumptions and were making improper and/or inaccurate changes to data. I didn't notice these systematic errors until the scope of the problem had gotten so large that it was having an effect on the collections-wide statistics I also gather every night (another tool I'd be happy to discuss or share, if you're interested). In an effort to discover these mistakes before they escalate to such a large scale, we've designed and implemented a python tool that is triggered by a nightly cron script. This tool (which we're provisionally calling the "CSpace change monitor") runs a series of simple, user-defined queries that each return a key-value pair for every record of a particular type (we've started with just collection objects, but will expand to other record types). The results of each of these queries are written to files each night, and then the current day's results are compared to the previous day's results. Any differences found are written to another file, and then these differences are run through some simple logic to filter out differences that have been deemed to be unremarkable (e.g., capitalization differences in Object name). The remaining differences are then formatted nicely and emailed off to the indicated recipient(s) for examination. Looking over the results takes less than a minute and any suspicious edits are quickly spotted so that they can be dealt with. Some planned or desired improvements: • make the report web-based and password-protected, with links to the records in question • include data about who made the changes (rather tricky, given CSpace's current audit capabilities) • enlarge the number of fields being checked (each field being checked adds about 20 seconds to the overall execution time, in a database with 650,000 object records). • enhance the logic for filtering out unremarkable changes Let me know if you're interested in getting more details about the tool and its code, or if you're interested in partnering up to work on improving this tool. Thanks, Michael Black Head, Research & Information Systems Phoebe A. Hearst Museum of Anthropology