Books & People

back Books & People Overview

Books, People

Archives

Book Reviews

 

Book Review: Introduction to Information Visualization

Book | Author | Review

By Gerd Waloszek, SAP AG, SAP User Experience – December 17, 2009

This review takes a personal look at Riccardo Mazza's book Introduction to Information Visualization.

 

Book

Cover of Introduction to Information Visualization     

Riccardo Mazza
Introduction to Information Visualization
 Springer, 2009
ISBN: 9781848002180

Information: Information presentation

 

Author

Photo of Riccardo MazzaRicardo Mazza is a lecturer and researcher at the University of Lugano, Switzerland and the University of Applied Sciences of Southern Switzerland (SUPSI), Department of Innovative technologies.

 

Review

Introduction

When I met Robert Spence briefly at the INTERACT 2009 conference in Uppsala, Sweden, he mentioned that there was a good book about visualization for display at the Springer book stand. I had already seen that there was "something" available about this topic – but actually it was two books: a new introductory book by Riccardo Mazza and a second, much more advanced one in its second edition by Chaomei Chen. The big question for me was: Which book had Robert Spence been referring to? I decided to buy both, and when I looked into them found that both referred to Robert Spence in one or another way. However, Chen seems to have cooperated with Spence sometime in the past. Anyway, I decided to leave my question unanswered, because the two books target different audiences: Chen's book addresses an advanced audience and even contains some mathematical formulas. Mazza's book, on the other hand, is an introductory textbook that "focuses on the human aspects of the process of visualization rather than the algorithmic or graphic design aspects." I decided that this was the book I should have read first when initially diving into the topic of visualization, but unfortunately it did not appear until this year.

While Mazza wrote his book "as a support text for a university course" (3rd and 4th year undergraduate and MSc students in computer science), he regards it as "also suitable for a wide and heterogeneous reading audience" that is concerned with communications systems on the basis of or predominantly using visual representations (practitioners). Thus, UI designers and HCI practitioners who want an easy introduction to the emerging field of information visualization are definitely included – hence my review.

A Few Definitions

Before the actual review, I would like to make a few things clear, based on Mazza's introductory chapter and beginning with the term visualization. According to Robert Spence, it denotes "solely a cognitive activity and has nothing to do with computers." While Mazza admits that some authors use the term visualization to refer to both the printed visual representation and the cognitive activity, he follows Spence and defines it as "a cognitive activity, facilitated by external visual representations from which people build an internal mental representation (mental model: added by the reviewer) of the world". Thus, in his book, Mazza "maintains the distinction between the creation of a pictorial representation of some data and the cognitive process that takes place when interpreting" it.

It may also be useful to recollect the distinction between data, information, and knowledge. Here, Mazza follows Shedroff: Data is "raw" entities lacking any meaning; these entities are transformed into information by processing, organizing, and presenting them, thus, by converting them into a meaningful form that human can be interpret. When information is integrated with existing experience, knowledge can be created, which should, according to Mazza, be the "principal aim of any communication process."

Finally, following Spence, Mazza distinguishes information visualization from scientific visualization. The latter is based on data having a correspondence to physical space (typically, it is closely related to mathematical models) such as geophysical data. Information visualization, on the other hand, refers to abstract data such as demographic or economic data that typically has no correspondence to physical space.

Overview of the Book

Mazza's book has eight chapters. It starts with a chapter entitled "Introduction to Visual Representations," in which the author defines "information visualization" (see above), shows its uses in explorative and informative data analysis, and provides criteria for good visual representations. He points to the issue of "visual" or "graphical" integrity: Using graphical techniques such as 3D effects and perspective for the sole sake of "coolness" bears the risks of distorting the relationships between data, directing attention to irrelevant information, and eliciting false interpretations. However, having such features at your fingertips in, say, Microsoft Excel, is probably much more influential than any warning in a textbook...

Chapter "Creating Visual Representations" begins by describing a reference model for the creation of visual representations. The model consists of three stages: (1) preprocessing and data transformation, (2) visual mapping, and (3) view creation. The author refers to this model throughout the book in order to indicate, which stage in the process a certain technique corresponds to. Then he outlines a procedure for designing a visual representation encompassing five steps:

  1. Define the problem or, better, the primary purpose of the representation (communication, exploration, confirmation of hypotheses)
  2. Determine the nature/scales of measurement of the data (qualitative, ordinal, categorical),
  3. Determine the number of dimensions/attributes (uni-, bi-, tri, or multivariate), and whether they are dependent or independent
  4. Determine the structure of the data (linear, temporal, spatial, hierarchical, network)
  5. Determine the type(s) of interaction on the data (static, transformable, manipulable).

I had some difficulties understanding how the model and the procedure relate to each other. Here is my interpretation: The reference model denotes the stages that data go through in a data visualization application. The central and most interesting stage is the mapping phase, where the numerical or categorical data is mapped onto graphical elements (points, lines, areas, shapes, etc.) and attributes (color, line length or width, etc.). The five-step procedure is meant to support a systematic design process for such an application, in particular for determining the details of the mapping (see also below).

Then Mazza goes into practice and demonstrates how linear data in one, two, and three dimensions can be represented visually.

The chapter on Perception steps back again, offering background information from human perception to help create effective and efficient data visualizations, thus to support the human mental activities that Mazza regards as "visualization." First, he lists the "classic" processing stages of human memory: sensory/iconic memory, short-term memory, long-term memory. Then he discusses preattentive properties, the most relevant of which for creating powerful visual representations are color, form, position in space, and movement. Preattentive properties can make data items "pop out" and are – according to Colin Ware (cited by Mazza) – "the most important contribution that vision science can make to data visualization." Mapping data to preattentive attributes is not an easy task and also has some limitations. Mazza offers a valuable table that helps in mapping/encoding data at different scales of measurement to various preattentive attributes (this table relates to step 2 in Mazza's abovementioned procedure). Then, he briefly discusses postattentive (conscious) processing and shows that it does not have any effect on preattentive processing. Mazza concludes this chapter by demonstrating basic gestalt principles (figure and ground, proximity, similarity, closure, continuity), unconscious innate mechanisms initially studied by German psychologists in the 1920s.

In the Multivariate Analysis chapter, Mazza returns to practice and linear data structures, this time, however, to those having more than three dimensions (multivariate data). He discusses the complexities of visualizing multivariate data and presents three approaches: geometric, icon, and pixel-oriented techniques. The chapter on Network and Hierarchies addresses another challenge, namely the representation of data with an underlying network or hierarchical structure, and lists a number of new and exemplary visualization approaches to both (BTW: mind maps are described in the section covering representations for networks; I had assumed that mind maps were primarily hierarchic). The chapter on the World Wide Web dwells further upon this subject and demonstrates how various Web-related data such as site maps, log data, search results, and interaction patterns in blogs or discussion forums can be represented visually to provide a better overview and new insights.

A chapter on Interactions addresses the final stage in Mazza's procedure for designing a visual representation: The author asks "How can users interact with visual representations in order to explore and understand complex data sets?" He distinguishes between manipulable and transformable representations: The former allow users to manipulate the resulting view, while the latter allow them to manipulate the data to be visualized, for example, through filtering or reordering. In most real-life applications, however, both kinds of interaction are provided. The final chapter in the book, Evaluations, offers a crash course through the field of human-computer interaction and the evaluation of visual representations using analytic and empirical methods. HCI professionals may read this chapter with mixed feelings, but Mazza's intention is, of course, laudable, namely to make the designers of visual representations aware of the need to verify their designs with users. The book closes with a list of references and an index. A list of figures is provided at the beginning.

Comments, Issues, and Suggestions

At first sight, Mazza's textbook comes close the ideal book for beginners in information visualization: It is – most of the time – easy to read, offers plenty of illustrations (I counted more than 90 in about 130 pages!), and comes in a handy, nearly pocketable format. Its about 130 pages do not put off readers – the book can be digested in a couple of hours. Reading the book, looking into the literature, and clicking the provided Web links takes, of course, a lot more time, but I find it a relief to see that a book can be – at least in principle – "swallowed" in one or two reads. (When I compare Mazza's book with the volume by Chaomei Chen mentioned above, I have to admit that the Chen book looks as if I would never manage to finish it.)

On the other hand, I stumbled across a couple of issues that in my eyes might impair the book's value for the target audience. Or put it the other way round: In my opinion, the book could be improved considerably if author and publisher would consider these ideas:

  • First, a book about visualization lives from the illustrations. While Mazza's book has a lot of illustrations, some of them are hard to read because of the choice of colors or because they are too small to recognize any details – even a magnifier did not help me much (I admit than I am no longer 20 and that my eyes have suffered over the years...). Sometimes, additional detail views are offered, but even these are sometimes too small. Of course, this shortcoming is also related to the small format of the book – you cannot have large illustrations in a small format. There is a suggestion for a dilution below.
  • My next gripe concerns the locations of illustrations and their corresponding text descriptions. When reading a description, you often have to scroll to another page to see the related illustration (and vice versa). I am aware that this is a tough issue for book authors, which has already been described by Colin Ware. He had tried to follow his principles in his book "Visual Thinking for Design" (Elsevier): "One of the (active vision) principles being that when text and images are related they should be placed in close proximity." – and he had had a hard time enforcing this. Mazza himself writes: "Information in short term memory is easily forgotten after a brief period of time" – this brief period of time could just be the time needed for turning one or more pages and searching for an illustration or description...
  • I admit that I have always difficulty in understanding text descriptions of complex processes or illustrations. This is the case for every book about information visualization that I have read (such as Robert Spence's book) and is also true for some of Mazza's descriptions. While close proximity would definitely help, a step-by-step description that isolates certain features of a visualization technique would help me a lot more. Such an approach would, however, require more space, adding further pages to the book and thus making it less handy... I would also like to suggest that authors check their explanations with members of their target audience in order to ensure that they are easy to understand.

All in all, the book remains far too much a "classical" academic book, and in my opinion, author and publisher missed some opportunities that would have made the book even more useful for its readers than it already is. Further reader support might include short sections listing related literature after each chapter and an overview of cited Websites (with the inherent issue that links may be broken or do not lead to the right information). My main proposal is, however, to follow the example of other textbooks and offer a companion Website to the book, providing larger (and more) images, extended descriptions, links to software and other relevant references, and more. This would definitely help address some of the issues mentioned above. I am aware that maintaining a companion Website means a lot of work, and many such Websites have been abandoned by their authors soon after the book had been published. But as Mazza is a lecturer for the topic of the book, I am sure he would keep it up for a while – and it would also support his lectures. By the way, on his homepage, Mazza offers a link to a wiki about information visualization that he maintains. I did not find any reference to it in the book, although I assume that most readers would find the Wiki useful. Perhaps, Mazza could transform his wiki into a kind of companion Website for the book.

Conclusion

Mazza's book definitely closes a gap by being – at least to my knowledge – the first introductory book to the relatively new field of information visualization. As already mentioned, introductory means also that the author "focuses on the human aspects of the process of visualization rather than the algorithmic or graphic design aspects". The book is not only – mostly – easy to read and understand, it also comes in a handy format that is ideally suited to reading anywhere. In addition, it offers plenty of illustrations, a mandatory requirement for a book on a visual topic. Despite a couple of issues, which, however, could be easily ironed out or at least amended in a new edition, I recommend the book not only to the primary target audience, students of computer science. I also recommend it to the intended wider audience, including UI designers and HCI practitioners, who are looking for an easy introduction to the emerging field of information visualization. After reading this introductory book they will not only be familiar with the relevant questions, approaches, and challenges regarding the field, they will also be able to much more easily step up to one of the more advanced books about information visualization, many of which you can find in our book list and in our special section about visualization.

 

top top