drupalib A place for library drupallers to hang out

I wanted to write up some of what we've got in the works just in case anyone has filled some of the missing pieces yet :)

Our virtual reference service records transcripts. The transcripts go in custom nodes. A custom index extracts URLs and another custom table extracts phrases from the patron's question. Phrases are groups of two or three words that don't contain a stopword and aren't broken by punctuation.

The slow, slow prototype, http://www.oregonlibraries.net/guides, shows popular phrases in questions in the past week. Clicking on a phrase (or searching for another) shows websites shared in transcripts where the question contained that phrase, sorted by how many times it happened.

For example

http://www.oregonlibraries.net/guides?crater_lake

This is a subject guide on the fly, broad or fine. Link checking is also done on the fly, so this loads slowly too.

As a next step, I have to trash my custom index.

* phrases go into a taxonomy
* URLs go into a CCK node type
* URL checking has to be done separately

And then we can Drupal the heck out of it. For example, here's a View of URL nodes tagged 'crater lake', with the Vote Up/Down Package module thrown in:

http://www.oregonlibraries.net/hottopic/crater_lake

Each time a URL shows up in a transcript, it goes into the CCK node. If a node exists for that URL already, it gets a vote.

Now librarians and patrons can vote on whether sites are good or not, or otherwise control the order that sites are listed.

I'd love to take this a step further and vote on whether or not a tag goes with a specific URL separately from whether the URL's overall score.

But I'd really like to expand this beyond websites. Libraries have books, too!

As a start, I used the Aggregation module to import an RSS feed of a search for Crater Lake in EBSCO's MasterFile premier, and vote on those as well.

http://www.oregonlibraries.net/hot_topic/crater_lake

The only sweat I put into this was theming the node layout for the views, and I see less sweat in our future if I can drop the custom module.

Also, setting up the EBSCO search was work, but so is answering chat reference questions and I think some of our librarians will be glad for a change of pace.

I think there are a couple of challenges here, including the future of aggregation in Drupal 6+, the need for a URL checker and a custom stopword module, but I'm happy the way this is shaping up.