Subject: Library Agents -Reply Date: Thu, 11 Jun 1998 10:11:47 -0500 From: "Gerry Mckiernan" Hi Travis Thanks so much for your interest in my posting. I appreciate your detailed response! <<< Travis Emmitt 6/11 9:13a >>> > > Gerry Mckiernan - > > Hello, I am doing IR research with Jim French at the University > of Virginia. I read your "Content Analysis of Academic Departmental > Homepages" message and am interested by your topic. I'm pleased to learn about your interest and the work at UVA. > I have some questions for you pertaining to your quest. First of > all, could you please give me some more specifics on what type of > responsibilities your envisioned Library Agents will have. More > specifically: > With such a collective user profile, one could now consider using > it as a Mega Search Statement that another agent would use to > search the Web > [snip] > Are you talking about tweaking the content of each user-supplied > query and then feeding the tweaked version to a run-of-the-mill > web searcher (pre-processing)? Or are you talking about > post-processing: filtering the results, or repackaging the results > back into a query, tweaking it, and sending it out again to another > (or the same) web searcher? [by "tweaking" I mean > tailoring/qualifying the query/results to maximize the chances of > obtaining all and only information "relevant" to the user, which > is in this case a department] My main interest here is to utilize the ***expressed*** subject interests of a departmental and that of their associated faculty embodied in the formal description of their research interests as the basis for a Mega search. Instead of searching the various sources individually or collective by the each of the faculty's stated interests, a Content Agent would search these departmental homepages and construct the appropriate Mega Search that would be passed on to Search Agents that would search previusly identified electronic databases of interests, e.g. publishers' catalog (and/or Amazon.com), e-journal sites of various journal vendors, the Online public catalogs of other libraries with relevant collections, to identify resources of potential interest to the librarian responsible for 'selecting' material for a local library collection. Here instead of manually performing the content analysis and manually searching the various electronic sources, the Agents would. [Would could imagine an adjunct within the Content agent that would message the results of the Content Analysis of the departmental/faculty homepages to include synonomous terms/concepts as the basis of an enhanced Mega Search statement. Here we would need to attached a Semantic Analyzer for this function. > If you were to envision, say, an Oceanography professor sitting > down at your Mega Search Engine, how would that session go in > your mind? How an end-user might interact with The System is open to many possibilities. In the very brief speculation of the Feedback Agent, my see the possibility of enhancing the formal ***expressed*** statement of research interests found in the homepage by adding/amending/integrating the results of a faculty member interaction with the results retrieved from the initial and subsequent Mega searches. [One could imagine creating a Research Interest Profile (RIP) based upon the ***Expressed*** interest in their individual departmental homepage. As they interact with the results retrieved from a Mega Search, their RIP would be tweeked to reflect any changes based upon this interaction [I would expect that the 'shape' of the RIP would change as feedback is received from the Feedback Agent, but that its overall shape would be within predictable semantic boundaries, e.g. a specialist in Fluid Dynamics would probably select items related directly or indirectly to Fluid Mechanics from the results of a Mega Search and not items that very indirectly related to his areas of focus, he probably would not (but might?) select items that dealt with the Mechanics of Body Fluids in Animals] > Now, another interpretation is that no human actual performs > the searches mentioned above, that a Library Agent is less a > searching assistant than an automatic "ferret" or "spider." > An LA would go out looking for relevant information and then, > when it finds it, would store or send it to the department/user > to which it is deemed relevant. > > In this case, the LA acts more like a proxy for a grad student > ("find me books about oceanography and then report back to me > when you've found at least 5") than an over-the-shoulder > librarian type ("could you please help me find books about > oceanography..."). Yes and No. The Agents would fetch but also 'recommend' [and there are plenty of Recommendation Agents] based upon the interaction behavior of the faculty member with the retrieved results. > Also, what exactly are you looking for: references to theory > and programs for Library Agents, or an actual Library Agent > that would help you with your work. Or both? Your references > to establishing a clearinghouse imply the former, but your > departmental home page references imply a need for an actual > working Library Agent. I am interested in the application of Agent technologies to library services and operations. Clearly, existing and emrging agents can be/could be/should be created for these applications. My posting is spectulation on one operation, e.g. Collection Development, and in this case utilizing the ***expressed** interests of the faculty member as the basis for this activity. > i.e., do you want someone to tell you about Library Agents > (you already know more than most about them) or do you want > someone to write you one? I am interested in promoting the application of Agents for Library Services and in stimulating thinking about these possibilities among my library colleagues. My hope is that reseaechers such as your team at UVA and elsewhere, as well as in commericial environments, might consider this application worthy of consideration for what I believe could be a very lucratiive market [I'll take my usual 10% [:-)] > Thanks! Thank _you_ for your interest and most of all for your time! Joy! Gerry McKiernan Science and Technology Librarian and Bibliographer Iowa State University and Curator, CyberStacks(sm) Iowa State University Ames IA 50011 gerrymck@iastate.edu http://www.public.iastate.edu/~CYBERSTACKS\