Position Paper

Barbara Mirel

Human Factors Manager, Lucent Technologies

My venture group within Lucent Technologies develops interactive information visualizations to help data analysts in business contexts solve such complex problems as how to reallocate resources or how to get a fair market share. The head developer, systems engineer and I ­ the head usability engineer -- are collaborating on a long term project to build a pattern language for visual querying and analysis. This visual query pattern language will first target problem-solving in the domain of retail category management and marketing.

Visual querying and analysis are new paradigms for inquiry, made possible by interactive, dynamically linked graphics, the technology that we as well as other software companies are developing for business problems. Our approach is to provide a software architecture that fosters complex problem solving while striving for interfaces that enable users to visually discover insights and draw conclusions in ways that are simple to master, efficient, accurate, and easy to learn.

Moving from familiar SQL statement or database/spreadsheet type querying to visual analysis requires competence in new methods and strategies of inquiry ­ for instance, accurately mapping data to the right graphic for a given question and purpose and directly manipulating graphically displayed data to drill down for detail, to transform variables, to merge tables, to compare and to relate parts and wholes.

Our groupís pattern building efforts are devoted to understanding and describing the entire process of visual analysis in a specific domain. We have developed prototypes for visual analysis in the retail category management and marketing domain and have conducted usability assessments to discover usersí processes of visual querying and analysis for key problems.  Our efforts are ongoing. Our goal is to integrate patterns from the object level to the higher level, purpose-driven activity level. Only in this way can we gauge whether design decisions are well-grounded in the actual uses that give analystsí problem-solving its vitality. From this integrated picture, we hope to design elegantly, making our product simple but not reducing the complexity of the usersí problems and decisions.

From our prototypes, usability assessments, and ongoing design experiences we have constructed an overarching, multi-layered framework for pattern building. At a user task level, visual querying  patterns include general as well as domain specific problem-solving patterns. At a graphic level, visual querying patterns encompass those related to visual perception and mapping problems to graphic representations. At an interactivity level, patterns relate to manipulating data directly as well as managing windows and inquiry processes. Finally, at a technology and performance level, patterns address the inner workings of objects for resusable components and application level solutions

Our progress so far includes building a hierarchy of general visual analysis and problem-solving patterns. These pertain to such aspects of querying as (1) Preparing and accessing data (13 patterns); (2) Planning inquiry (8 patterns); Querying-and Interpreting (33 patterns); Communicating/Reporting (12 patterns); and Managing Inquiry (16 patterns).  We also have built a beginning set of patterns for some of the most important and frequently performed problem-solving goals and subgoals in category management and marketing. As we learn more from our usability and design efforts, we modify our patterns accordingly.

In the rest of this proposal I first describe a group of patterns relevant to one QUERY-AND-INTERPRET subgoal in category management and then as an example I detail a single pattern within this larger story.

A position paper prepared for the Usability Pattern Language Workshop at Interact '99