Despite the advent of graphical user interfaces about thirty years ago at Xerox PARC, most catalogues are built on text. Libraries had their origins in text, books and other written documents and, in many cases, that is what our institutions are still known for (despite the fact that libraries do much more than that. The focus on books and text in our collections has also influenced the design of our catalogues. From my unscientific sample of major and minor library catalogues, I find that almost all catalogues are fundamentally drive by text: all the fields are in text, input is text and all the results are textual. Don’t get me wrong – I love text. As Thomas Jefferson remarked, “I cannot live without books.” Reading and writing is my preferred learning style (though I very much appreciate a well crafted lecture) but I recognize that other styles exist out there and it is important for me to learn about them even if they don’t feel natural.
In my understanding of the concept, visual search can mean at least two different results: searching for and/or displaying images or displaying search results using visuals that go beyond lines of text. Although some researchers are working on searching images directly, most systems I am familiar with (e.g. Google Image Search) use meta-data such as HTML tags or catalogue details to locate images. Such searching on Flickr can often produce interesting ideas and amusements though the results are uneven in quality. The other type of visual search is more about interface – showing results to the user such as tag clouds (e.g. tag cloud of the books at LibraryThing) or other methods. Of the two, I think that libraries are more likely to be interested in working on visual display of search results but the other type is certainly still of interest, particularly as large collections of photographs are digitized.
Based on experimentation with a few different Web search engines (are there library catalogues using it out there? If so, please let me know in the comments), my impression is this feature is not ready for prime time. Of all tools I tried, I found Spezify to be the most fun if quirky (it mashes together tweets, with images and webpages; doing a search on a city is fun). This kind of search is still only of interest to those who enjoy experimenting with such tools – I don’t know that it could be used to find something specific as yet. The experience of visual search could best be described as akin to browsing the book stacks of a library; accidental discovery is the order of the day but not much more.
Has anyone out there found a really good visual search tool? How does it work? I have a feeling I may have missed some good examples of visual search – would anyone care to correct me with a stellar example?