Complex data made easier on the eyes

Information Visualization is a new quarterly journal that deals with computing systems capable of displaying complex quantitative data, such as weather charts, stylised underground railway maps and graphics embedded in financial news items. In our multimodal age, the subject is not only popular but is also of strategic importance for a wide range of users - from global enterprises seeking to interrelate the preferences of their customers in different locations, to academics wishing to display the results of an experiment or a computation that involves large data sets.

More recently, information visualisation techniques have been developed to deter terrorism. Information visualisation, a new branch of computing (c. 1990s), links human computer interaction, computer graphics, cluster analysis, neural networks and many other subjects.

The cognitive act of visualisation relates to the formation of a mental vision, image or picture of something. In computing, visualisation is defined as "the display of data with the aim of maximising comprehension rather than photographic realism". The question addressed by Information Visualization is whether or not computing systems can enable humans to visualise better and, in the longer term, whether such systems can learn to visualise like humans.

The journal's editorial board includes leading figures in the subject such as Ben Shneiderman and Robert Spence. In the first of its papers, Shneiderman notes that information visualisation encourages a realistic and empirical approach to matters scientific. In particular, it is about displaying complex data in as much detail as possible without distracting the viewer. This empirical approach dates back to historians and chartmakers of the 18th and 19th centuries such as William Playfair and Charles Minard. However, it was overshadowed by the rise of rationalism in the late 19th century and early 20th. For rationalists, observations had to be summarised into statistical metrices - means, standard deviations, principal components, data clusters and so on - acting as surrogates of the original data. "Statistical thinking" is just as important as "visual thinking", and a balance has to be struck between the two to reach an informed and precise conclusion.

Information visualisation has a symbiotic relationship with data mining - another new branch of computing that focuses on the analysis of large volumes of data captured, for example, through loyalty cards used by supermarket customers. Data mining is dominated by statistical analysis (or "statistical thinking"), including multivariate statistics, cluster analysis and machine learning. It subscribes to the rationalist approach, while information visualisation subscribes to the empiricist approach. The two complement each other and there may be advantages in using both in the analysis and display of complex data.

Papers published to date in Information Visualization include topics such as how to display large volumes of multidimensional data using parallel coordinates; what the possible successors of the renowned/reviled PowerPoint might be; and data fusion. User evaluation appears prominently in many articles.

These are early days for Information Visualization . The papers are informative, with regular contributions from leading figures in the field.

Contributors are drawn from academia as well as computing enterprises and, while coverage mostly focuses on visualisation, the journal includes issues related to statistical analysis, neural computing and occasionally data mining.

Academic competition comes from the ten-year-old quarterly journal, Institute of Electrical and Electronics Engineers Transactions on Visualisation and Computer Graphics , which published more than 130 peer-reviewed papers between 2002-04, compared with 50 or so in Information Visualization over the same period. However, despite the word "visualisation" in the title, the papers in Transactions deal largely with computer graphics - a subject dedicated to the pictorial representation of objects and data - so it does not pose serious competition to Information Visualization in the short term. Palgrave Macmillan is, therefore, providing a valuable service by publishing this newer journal and, like any publication - especially one without institutional muscle - this one needs tender loving care. Subscriptions will start to flow as soon as researchers become aware that there is one journal that encourages an interdisciplinary approach to the visual display of complex quantitative information.

I have just two criticisms. First, the quality and layout of graphics: the colour plates sometimes lack zest, and better quality could justifiably be expected of a journal on visualisation. Figures used to illustrate papers are seldom on the page where they are referred to, which is certainly not good visual display of information. The second relates to the added value that editorials could give to the journal. Signed editorials are a regular feature, but these are in essence a summary of content, which, while helpful, does not constitute an editorial. Sometimes these editorials are puzzling. For example, the one to celebrate the first full year of publication is titled "Information visualisation is growing" and uses data to support such a conclusion partly drawn from a "crude search query" to a well-known citations database; the graphics used are also somewhat confusing. These criticisms notwithstanding, Information Visualization has made a good start and I look forward to future issues.

Khurshid Ahmad is professor of artificial intelligence, Surrey University.