HomeFInder: Screenshot of one of the first versions with the sliders to select the range of variables and a map of Washington DC with the available homes.: Source: HCIL, University of Maryland. Click on the image to enlarge it

Spotfire has its ancestors in a system that experimented with the concept of dynamic queries, a little more than a decade ago. Said concept, first reported by Christopher Williamson and Ben Shneiderman in 1992*, was applied at that time to the search for homes appropriate to the requirements of their potential buyers or tenants. 

In the words of Bob Spence in his book “Information Visualization” the problem a dynamic query solves is 

“Given a collection of objects, each described by the values associated with a set of attributes, find the most acceptable such object or, perhaps, a small number of candidate objects worthy of more detailed consideration.”

In the initial example of Williamson and Shnider’s article, the set of objects was the contents of a real estate database, with thousands of homes for sale or rent; the attributes were the features of every house, square feet, number of rooms, terrace, garden, price, location, etc.

At that time (like it still is today) the process of querying a database searching for a house with at least four bedrooms, two bathrooms for less than 120.000 euros could be done using SQL, more or less like this

SELECT house FROM real_estate_database WHERE price <= 120000 AND bath = 2 AND rooms >= 4

The answer could be 0 addresses (prices are skyrocketing) or a large number of results depending on the country or region considered. In any case the results were not a huge help and required an operator that knew the query language of the database, usually intolerant to syntax errors. Moreover the results (a list of addresses) hide context information that could be useful for the buyer. 

The solution to this problem is to realise that in many cases the users formulate a problem while at the same time try to solve it. As we are dealing with the problem and receive more information we find new ideas for our house or we discover that we don’t actually need 4 rooms. In order to take this into account the dynamic queries continuously update the queries, which are done visually by means of manipulating sliders and scroll bars (see the image in the graphical version of this article), applying the principles of direct manipulation to the database:

• visual presentation of the components of the query (typically with sliders and scroll bars) 
  
  
• visual presentation of the results (application dependent, uses the usual techniques for graphic representations).
  
  
• rapid, incremental and reversible query control.
  
  
• selection by pointing, not by typing in the command line.
  
  
• continuous and immediate feedback.

A further and more elaborated application of these principles led to an interesting application, FilmFinder, result of the work of Chris Ahlberg and several students, that showed more than 1500 films classified by their genre (drama, comedy, horror…), popularity, duration, author, etc and allowed the user to select by means of sliders (called alphasliders) which films appeared or disappeared according to the ranges the alphasliders allowed him/her to define.

FilmFinder: A particular selection of a film, showing the available information of the database. The image of one of the actresses has been selected. Source: HCIL, University of Maryland  Click on the image to enlarge it	FilmFinder: The selection shows all the available films. You can see the volume of old films is scarce while you have many recent ones. Every coloured square is clickable showing information about the movie. Source: HCIL, University of Maryland  Click on the image to enlarge it

As Shneiderman explains in “Dynamic queries, starfield displays, and the path to Spotfire” once Ahlberg finished his PhD and developed a more commercial version of the software, he gathered together some friends and venture funding to create a company that was called IVEE Development that launched the product in 1996 with the name of Spotfire. The company itself eventually changed its name to Spotfire

AlphaSlider: Unlike scrollbars these type of sliders allow you to select a range of values of a specific variable. Only the objects that correspond to that range will be displayed. Inthe lower part the possible values are shown. The upper part show the selected range. Source: HCIL University of Maryland

I’m interested in three aspects of Spotfire:

• the powerful advance that the concept of dynamic query represents, that has spread to many other systems like CityO’Scope
  
  
• the “alphaslider”, a slider that allows you to select ranges (more than x but less than y). 
  
  
• the fact that Spotfire has survived as a company with already 8 years of history.

Spotfire has become one of the (still) few pioneering companies that make a living from information visualisation, which is specially meritorious in such an emergent field where there’s still little awareness about the potential of its use.

You still can see HomeFinder in action

At number 128 we also spoke about dynamic queries.

* “The dynamic HomeFinder: evaluating dynamic queries in a real-estate information exploration system” C. Williamson & B. Shneiderman, 1992 ACM, Preceedings SIGIR'92 pp 339-346

Links of this issue: