it could be worse

Have you noticed the changes Google has been making to the way they display search results? Google Instant has been the latest, but before that, there was the introduction of the “Everything” sidebar. And that one in particular seems to have upset numerous Google search fans. If you do a search in Google for “everything sidebar,” the first few results are about removing or hiding it.

Not only that, but the latest offering from the Funny Music Project is a song all about hating the Google “Everything” sidebar. The creator, Jesse Smith, expresses a frustration that many of us can identify with, “It’s hard to find a product that does what it does really well. In a world of mediocrity, it’s the exception that excels. Then some jerk has to justify his job by tinkering and jiggering and messing up the whole thing.”

Tech folks like to tinker. We like making things work better, or faster, or be more intuitive. I’ll bet that there are a lot of Google users who didn’t know about the different kinds of content-specific searches that Google offered, or had never used the advanced search tools. And they’re probably happy with the introduction of the “Everything” sidebar.

But there’s another group of folks who are evidently very unhappy with it. Some say it takes up too much room on the screen, that it adds complexity, and that they just don’t like the way it looks.

Cue ironic chuckling from me.

Let’s compare the Google search results screen with search results from a few of the major players in libraryland:

Google

ProQuest EBSCOhost

CSA Illumina ISI Web of Knowledge

So, who’s going to write a song about how much they hate <insert library database platform of choice>?

ER&L 2010: Where are we headed? Tools & Technologies for the future

Speakers: Ross Singer & Andrew Nagy

Software as a service saves the institution time and money because the infrastructure is hosted and maintained by someone else. Computing has gone from centralized, mainframe processing to an even mix of personal computers on an networked enterprise to once again a very centralized environment with cloud applications and thin clients.

Library resource discovery is, to a certain extent, already in the cloud. We use online databases and open web search, WorldCat, and next gen catalog interfaces. The next gen catalog places the focus on the institution’s resources, but it’s not the complete solution. (People see a search box and they want to run queries on it – doesn’t matter where it is or what it is.) The next gen catalog is only providing access to local resources, and while it looks like modern interfaces, the back end is still old-school library indexing that doesn’t work well with keyword searching.

Web-scale discovery is a one-stop shop that provides increased access, enhances research, and provides and increase ROI for the library. Our users don’t use Google because it’s Google, they use it because it’s simple, easy, and fast.

How do we make our data relevant when administration doesn’t think what we do is as important anymore? Linked data might be one solution. Unfortunately, we don’t do that very well. We are really good at identifying things but bad at linking them.

If every component of a record is given identifiers, it’s possible to generate all sorts of combinations and displays and search results via linking the identifiers together. RDF provides a framework for this.

Also, once we start using common identifiers, then we can pull in data from other sources to increase the richness of our metadata. Mashups FTW!

CiL 2008: Speed Searching

Speaker: Greg Notess

His talk summarizes points from his Computers in Libraries articles on the same topic, so go find them if you want more details than what I provide.

It takes time to find the right query/database, and to determine the best terminology to use in order to find what you are seeking. Keystroke economy makes searching faster, like the old OCLC FirstSearch 3-2-2-1 searching. Web searching relevancy is optimized by using only a few unique words rather than long queries. Do spell checking through a web search and then take that back into a reference database. Search suggestions on major search engines help with the spelling problem, and the suggestions are ranked based on the frequency with which they are searched, but they require you to type slowly to use them effectively and increase your search speed. Copy and paste can be enhanced through browser plugins or bookmarklets that allow for searching based on selected text.

The search terms matter. Depending on the source, average query length searches using unique terms perform better over common search terms or long queries. Use multiple databases because it’s fun, you’re a librarian, and there is a lack of overlap between data sources.

Search switching is not good for quick look-ups, but it can be helpful with hard to find answers that require in-depth query. We have a sense that federated searching should be able to do this, but some resources are better searched in their native interfaces in order to find relevant sources. There are several sites that make it easy to switch between web search engines using the same query, including a nifty site that will allow you to easily switch between the various satellite mapping sources for any location you choose.

I must install the Customize Google Firefox plugin. (It’s also available for IE7, but why would you want to use IE7, anyway?)

CiL 2008: Catalog Effectiveness

Speaker: Rebekah Kilzer

The Ohio State University Libraries have used Google Analytics for assessing the use of the OPAC. It’s free for sites up to five million page views per month — OSU has 1-2 million page views per month. Libraries would want to use this because most integrated library systems offer little in the way of use statistics, and what they do have isn’t very… useful. You will need to add some code that will display on all OPAC pages.

Getting details about how users interact with your catalog can help with making decisions about enhancements. For example, knowing how many dial-up users interact with the site could determine whether or not you want to develop style sheets specifically for them, for example. You can also track what links are being followed, which can contribute to discussions on link real estate.

There are several libraries that are mashing up Google Analytics information with other Google tools.


Speakers: Cathy Weng and Jia Mi

The OPAC is a data-centered, card-catalog retrieval system that is good for finding known items, but not so good as an information discovery tool. It’s designed for librarians, not users. Librarian’s perceptions of users (forgetful, impatient) prevents librarians from recognizing changes in user behavior and ineffective OPAC design.

In order to see how current academic libraries represent and use OPAC systems, they studied 123 ARL libraries’ public interfaces and search capabilities as well as their bibliographic displays. In the search options, two-thirds of libraries use “keyword” as the default and the other third use “title.” The study also looked at whether or not the keyword search was a true keyword search with an automatic “and” or if the search was treated as a phrase. Few libraries used relevancy ranking as the default search results sorting.

There are some great disparities in OPAC quality. Search terms and search boxes are not retained on the results page, post-search limit functions are not always readily available, item status are not available on search results page, and the search keywords are not highlighted. These are things that the most popular non-library search engines do, which is what our users expect the library OPAC to do.

Display labels are MARC mapping, not intuitive. Some labels are suitable for certain types of materials but not all (proper name labels for items that are “authored” by conferences). They are potentially confusing (LCSH & MeSH) and occasionally inaccurate. The study found that there were varying levels of effort put to making the labels more user-friendly and not full of library jargon.

In addition to label displays, OPACs also suffer from the way the records are displayed. The order of bibliographic elements effect how users find relevant information to determine whether or not the item found is what they need.

There are three factors that contribute to the problem of the OPAC: system limitations, libraries not exploiting full functionality of ILS, and MARC standards are not well suited to online bibliographic display. We want a system that doesn’t need to be taught, that trusts users as co-developers, and we want to maximize and creatively utilize the system’s functionality.

The presentation gave great examples of why the OPAC sucks, but few concrete examples of solutions beyond the lipstick-on-a-pig catalog overlay products. I would have liked to have a list of suggestions for label names, record display, etc., since we were given examples of what doesn’t work or is confusing.