This is the first Access Conference I've attended, but it was well worth the effort to be there. This year, the conference was held in Canada's capital, Ottawa, in the province of Ontario. The focus of this year's conference was "capitalizing" on work that has been done and look at future initiatives and excellent projects that are currently in development in the library technology field. Several professionals from information and library related fields were present at the conference, and gave excellent presentations. A complete recap of the conference is available at the official Access 2006 web site, but I'll take a look at some important highlights here.
Art Rhyno and Walter Lewis spoke briefly about an exciting project they are pursuing called "OurOntario" which is a collection of diverse digital objects such as images, books, newspapers and artifacts about Ontario and the history of Ontario. One of the important "hot topics" at this conference was indexing, and Lucene took center stage. As both Art and Walter pointed out, Lucene is a very flexible and robust indexing application which is entirely written in Java, and requires a Java API in order to use. Lucene has a full range of options for indexing content and allows the ability to merge heterogeneous indexes together into a master index, so large indexing projects such as the OurOntario project can be load balanced across several commodity machines and significantly increase the speed of indexing operations. Since much of the content of OurOntario is of a historical nature, it has been difficult to find accurate and useful information about, for example, old photos from the Ontario area. In order to try to increase the reliability of the metadata, and to make it more authoritative, Walter has created the OurOntario interface in such a way that it allows for generous amounts of "feedback" and "user input" regarding the digital objects themselves. So, for each object represented in the application, a blog-like "comments" tab is available in order to allow users to give feedback to the project coordinators. The ultimate effect of this is that the metadata maintainers have the opportunity to correct and enhance the information about each object, and the feedback is attached to the digital object itself for everyone to see. This method of updating metadata is very creative and has a sort of "wikipedia" feel to it, and adds a layer of richness to the content.
We've been hearing an increasing amount of talk these days about the "next generation OPAC". I think it is universally agreed upon that current discovery tools are broken in comparison to such commercial sites as Google and Amazon. NCSU Libraries has recognized the crisis, and decided to take on the challenge with Endeca. Tito Sierra, a former employee with Amazon, gave a very interesting presentation on the NCSU Endeca implementation. He said that the primary motivation for this move was to enhance the quality of the user experience for NCSU library patrons. Some of the qualities they were looking for in an alternative to the status quo include: relevance ranking, enhanced browse functionality, intuitive subject access and performance/speed. They also wanted to include "search comforts" such as a 'did you mean' service, thesaurus functionality, noun stemming and specialized sorting such as "rank by popularity". They can take advantage of the latter by exploiting circulation data. The way Endeca works at NCSU is via several stages: a. An export process runs their records through a "foundry" to normalize the records, b. An indexer runs which creates high speed indexes and "mdex" compatible records c. Endeca uses these indexed records to allow browsing, create discovery tools and build services. Browsing is accomplished by taking advantage of authority records. Endeca is a back end service, requiring libraries to build applications which will be public facing, and NCSU obviously has the staff and skills to accomplish that. Overall, they have put together a nice package using a flexible tool.
The lightning talks, which are limited very strictly to 5 minutes each, allowed several folks to talk about and demonstrate some great projects that are going on. In particular, Jeremy Frumkin from Oregon State University, had a chance to briefly talk about their metasearch project which looks very interesting. The project looks promising for many reasons, but it seems as though the OSU developers are taking pains to incorporate usability testing, they are trying to create an easy to use results list, and they are also trying to take full advantage of AJAX technology to simplify the user experience. Another great but brief talk was about the MarcEdit program, also hosted at OSU. This looks like a pretty slick program with a lot of standards based features that should be helpful for developers.
The "thunder" talks were filled with all sorts of goodies. Everything from Ross Singer's "ümlaut" presentation regarding an open-url project they are working on in an attempt to make the OPAC display as full of information as possible without crowding the display with useless information, to Dan Chudnov's rapid fire update on several projects he is either helping with or has initiated such as COins, unAPI , and his project about interviewing "geeks" and giving geek tips. One of the really interesting thunder presentations highlighted a web services layer that runs on top of Lucene called SOLR. SOLR is another Apache Foundation project which provides a "web services - like API" with features such as full featured advanced search APIs, completely standards based open interfaces that can return XML or HTML, filtering and replication mechanisms, and an administrative interface. I haven't investigated this further, but it may provide a good mechanism for interfacing with other open source languages such as Ruby, Perl, PHP, etc with Lucene. By far, though, my favorite thunder talk was titled "Get Over It, Fanboy" presented by Mark Leggott. This was a plea to library and institutional administrators to encourage exploration in technology in order to build skill sets and interest levels. It was a fun presentation, and I hope it helped folks who haven't taken the time to experiment with technologies that they have never worked with to dive into the deep end. In order to succeed, we need to become a "fan boy" of a particular technology and possibly convince administrators that our new, favorite tool is the right tool to get the job done. Of course, this requires effort and the benevolence of departmental supervisors and institutional administrators. Hooray for fan boys!
Virginia Tech Libraries developers have implemented an extension for Firefox (unfortunately not yet for IE, but that is in development) which allows library patrons to utilize library resources which include OPACs, databases, indexes, and interfaces with OpenURL resolvers such as SFX, all without needing to navigate to the library's web site. This application is called LibX. Included in the package is a toolbar, a right click menu and a web localization component which can add links to web pages that the user visits such as Google. LibX is completely open source and is becoming very popular. Several important academic libraries (such as Harvard and MIT, I believe), have contributed to the project and have adopted it. In the demonstration that I saw, several nice features such as ISBN searches taken from Amazon were incorporated so that with one click the user could get information directly from the Virginia Tech online catalog. They were also able to incorporate the VT logo at the top of Google searches to easily allow users to find out if what they are looking for is possibly contained in the local academic catalog.
The University Library of Hamburg has been hard at work perfecting a library "chatbot" called Stella. The woman who gave this presentation said that when they first implemented Stella, the "success rate" of being able to direct patrons to the appropriate resource based on their interaction with the chatbot was between 60-80%. They now rate their success at somewhere around 85-90%. The chatbot is a natural language "artificial intelligence", although the woman presenting denied that actual artificial intelligence is involved, that can direct patrons or answer their questions based on what they type into a query box, dynamically. The responses from Stella are based upon a large database of FAQ like information that has been collected over time. The chatbot is available 24/7 and has increased the overall usage of e-journal, database and full text content at the University. Apparently the chatbot is life-like enough that she has received several marriage proposals.
Stan Ruecker, the winner of the David Binkley Emerging Technology Award, gave a presentation about an experimental method he, and various teams of researchers at the University of Alberta, have been working on in order to visually represent and otherwise abstract set of data / metadata. For example, Stan demonstrated a set of tools that can visually organize data according to an arbitrary set of characteristics which have been previously evaluated such as the overall mood of a poem or the nature of relationships between characters in a story. The arbitrary elements can then be visually grouped in order to represent information in a more intuitive context. I think the idea is that a picture is worth a thousand words, and if you have a very large set of related data, it can be interpreted very rapidly and meaningfully in a visual context.
Thom Hickey, from OCLC, gave a presentation on an authority control project they are working on called VIAF, the virtual international authority control file. This project was jointly initiated by OCLC, the Deutche Bibliothek, and the Library of Congress. Currently this project is processing over 19 million "identities" comprising 80 million controlled headings. Essentially, the idea is to produce a very robust authority database which has been enhanced with useful data, and can be called using web services techniques and joins togrther international efforts at authority control. Also, the joint database will be able to syndicate authority information which is appropriate to the national language where the request is made (for example, German in German speaking countries, etc). Thom went through, in a very detailed fashion, how the identities are created, which involves multiple stages including queries to WorldCat, the incorporation of FRBR data, cover art for the related works, and an audience designation. The OCLC authority index is also consulted along with references to Wikipedia. Once all of this information has been consolidated, an identity is produced which is represented in the form of an XML record. Currently, this project is in an experimental stage and a firm date has not been released when this will be available to OCLC customers.
On the second day of the conference, the well known Roy Tennant gave a talk on what he called "the last mile", which for him refers to getting patrons that last mile to what they need. A catch phrase for his presentation was "two clicks away"...one click for the search during the discovery process, and one click to the resource. This is the ideal. Roy demonstrated that, even at his own library, we sometimes create artificial or unnecessary barriers to content. For example, he discussed a test case where a patron searches for a book in a catalog, and while the catalog is available in the library, the patron is directed to Google Books where only parts of the item are available online. To make matters worse, the full text of the monograph is actually available if one searches the standard Google interface. In this scenario, we have obfuscated the resource instead of making it easier for the patron to acquire it. In may instances, Roy blamed OpenURL as the culprit, because it creates another artificial layer between the patron and the content, and at times can mislead the patron into believing that they need to use ILL when the item is actually available in the library or available online in full text format. However, even given these problems, Roy gave examples that are attempts to fix discovery systems that more easily connect patrons to resources. However, as he pointed out, while OpenURL resolvers are a good start, they are not a panacea, nor are we using them to their best potential. He advocated that instead of forcing patrons to use yet another web interface through which they must click, incorporate those services directly into the OPAC interface. Web services can be used to incorporate links to content or links to services on one interface. While this is not an easy problem to solve, there is hope.
Overall, this was a very informative conference that helped to get the creative juices flowing, and also confirmed for me that we here at the University of Notre Dame are going in the right direction as we try to integrate open standards, web services, and creative indexing into our suite of patron services. There's always so much work to be done, but this conference demonstrated what can be done, and what is being done. I sincerely hope that we can try to move forward with technologies like web services, AJAX, Lucene and OpenURL, pushing the technologies to their greatest potential, and that we collaborate more with other institutions where similar projects are being done.
All libraries:
Architecture | Art
Image | Business Information Center
| Chemistry & Physics
| Engineering | Hesburgh
(Main)
Kellogg/Kroc Information Center |
Life Sciences | Mathematics
| Rare
Books & Special Collections | Radiation
Lab | Kresge Law
