There is much discussion going on in semantic technology circles about how to determine authority. It has reminded me of my Reference Services course with A.J. Anderson at Simmons College GSLIS.

When I was in library school we were taught how to ascertain the authority and provenance of our sources, to ensure we were providing our patrons with the most trusted sources of information we could find. It was about RESPECT and HONOR; about doing the right thing. It was not about algorithms. No mathematically derived number will match the passion of a dedicated human, intent on finding the best source of information available with the resources at hand, with the mindset that s/he is there to provide impeccable SERVICE and if possible EDUCATE the patron.

That being said, how can we define algorithms that let us use the machine to support the human's effort? I look for integrity in my sources, but define its attributes a little differently than others in computer science fields I think.

1. Is the source complete - whole/intact?
2. Is the source broad and shallow (an overview) or narrow and deep (comprehensive coverage of a single or few topics.)
3. Is the source there when I need it? (On the web - is the URI persistent/no "Page Not Found" errors. In print - this is why libraries have reference sections from which you cannot borrow!)
4. Is the source factual - research-based, critical, op/ed?
5. Is the author of the information well-suited to present on the topic? Degreed? Real-world experience of any tenure? How much experience? Participates in open discussions on the subject? Regularly published in the same or related fields? Highly regarded by peers qualified to make such a judgment? Unbiased by funding sources?
6. Does the source reference other materials of the same AND differing research results or opinions?

Frequently though, despite all of this effort, at the end of a search I will find myself selecting the "right" source much the same way as the experts in Malcolm Gladwell's Blink did - I trust my instincts. Why? Because I don't often have all of the data I need to determine if a source passes the above tests.

I'd like to see an app that indexes and visualizes these elements for me - an app that visualizes what each individual determines is their own definition of integrity. Perhaps having good reviews by others of a similar knowledge/experience level as themselves on a topic is important. Perhaps having the opinions of friends is important.

I'd like an app in which each of these components is plug-n-play, not pre-determined for me like Google's 200 part algorithm.

Of course, I'll make the effort to use it.

In the meantime, please tell me in the comments what matters to you when determining authority, trust or integrity of an information source. I'm genuinely curious to learn and want to help move the dialog along.

[Interestingly, this Google search provides some excellent information on how librarians recommend you determine authority. Check them out and see if you can determine if these resources have integrity!]

I am frequently asked to explain some of the more abstract concepts in taxonomy and ontology work. I thought I'd start writing them down as I come up with them. Feel free to comment on their success - I hope some people find them useful!

Deprecation is a very useful technique that is often overlooked. In the context of taxonomies and ontologies it means to stop using an old term in favor of a new one. Rather than simply deleting the old term, creating the new term and then emailing everyone you think cares with a note about the change, you can include it in your model. This is a much more thorough way of ensuring all of the people and applications that use your data have the up-to-date information.

Some examples: search engines can expand their given query to cover both terms; indexers can be given the most recent version to tag with; end users can be notified via an RSS feed of the change.

It's not as strange a technique as you might think. We encounter deprecation regularly. The image to the right shows one example. This is a road sign in Pennsylvania. It used to be that highway exits were numbered sequentially from one end of the highway to the other. Of course, suburban sprawl demanded the addition of off- and on-ramps, and new exit numbers had to be slipped in, leading to signs reading "225a," "225b," "225c" and so forth. Many transportation departments decided to simply renumber all of the exits based on the nearest mile marker, such that exit 37 is within the 37th mile of the highway from it's beginning measurement point. To help ease confusion for the driver and/or navigator, the old exit was noted on the sign post as well as the new. This is deprecation. This sign is saying, "If you want Street Road, take this exit, number 37. You may know it as exit 25, but we changed the number. Please use this new number from now on."

Deprecation - reflect the new, respect the past.