Today was a good day, so I thought I would share its results immediately, instead of fine-tuning forever — who knows when I find the time anyways!

I built a little facet browser for the New York Times Article Search API - an impressively fast faceted search engine covering over two million articles. So, give it a spin!

Some caveats:

• Don’t look for the page navigation — there is none. Pure laziness, will update it soon.
• The initial counts are based on a search for “the” (which I figured would appear in all articles). Unfortunately, only the top 15 or so values per facet are returned, so you cannot click, e.g. the year 2008 in the beginning. Will fix.
• The API has a request limit of 5000 queries per day. So if your requests don’t work — come back tomorrow morning :)
• Unfortunately, the API seems to support only one value per facet. So, all facets are single-select.(fixed, see comments).

The code is based on my totally revamped elastic lists prototype. I used this project as a little sandbox experiment of how easy customization is possible, and especially how to make a switch from a fully client-based to a server–based filtering model.

A real wow-project has gone into version 2: Exhibit. It is part of SIMILE, focussing on “Semantic Interoperability of Metadata and Information in unLike Environments”, which provides a whole toolbox of pragmatic semantic web applications.

Exhibit itself presents a “a three-tier web application framework written in Javascript, which you can include like you would include Google Maps.” An exhibit application typically consists of a content presentation area and several widgets for filtering, sorting and grouping the content presentation, following the faceted browsing paradigm also used in my elastic lists. Currently, the available widgets cover plain checkbox lists, maps, and timelines, and a live search.

From my first experiments with the tool, I can say it is really ridiculously easy to create custom views on existing data sets: Data can be imported “from a Google Spreadsheet, Excel spreadsheet, EditGrid spreadsheet, BibTex files or any JSONP data source” without much effort.

Applications are built by loading the exhibit script and then assigning “Exhibit roles” to layers in your java-script: For instance,

<div ex:role="facet" ex:expression=".discipline"
ex:facetLabel="Discipline" />

will create a filtering widget looking for values of the “discipline” property in your data set and display them as a list to be used for filtering. It couldn’t possibly be easier to build client-side facet browsing applications.

Other nifty are the integrated browser history for filtering steps and the ability to export filtered views as e.g. HTML, but also tab-separated text files, RDF, etc.

Check out the Getting started page to get an impression.

The down side, of course is, that all data has to be loaded on the client before the application can start. So we are rather speaking about hundreds than thousands of data items. Also, I would of course be interested in building custom widgets, however, this looks a wee bit more complicated from first looks into the code. And of course UI-wise, I would have some suggestions as well, but anyways, I find the project quite impressive already as it is!

Kudos to David Huynh and the rest of the team at MIT. More info can also be found in David Huynh’s PhD thesis.

It is a demonstration of the “elastic list” principle for browsing multi-facetted data structures. Click any number of list entries to query the database for a combination of the selected attributes. If you create an “impossible” configuration, your selection will be reduced until a match is possible.

The example data is based on the Noble prize winners dataset used in the Flamenco facet browser.

Elastic lists enhance traditional facet browsing approaches by • visualizing relative proportions (weights) of metadata values by size • visualizing unusualness of a metadata weight by brightness • and animated filtering transitions.

In unfiltered view, the brightness shows a trend measure, indicating a rising number of prices of the last years.

In filtered views, a brighter background indicates a higher weight of the metadata value compared to the overall distribution.

If, for instance, you click “peace” as in the example above, you will see that “female” and “Switzerland” are much brighter, indicating that the proportion of women and Swiss is much higher in this context than compared to the whole data set. That’s interesting information and could also be used to characterize the result set of a keyword query or any other collection in terms of its “characteristic” metadata values. Besides that, it fosters understanding of how metadata values are correlated with each other, which is often interesting information itself.

You can also switch on little sparklines to see the temporal distribution of each metadata value: