Design Patterns for Spatial Information Visualisation and Analytics Applications

Earlier this week we added another four patterns and two topic pages to the growing collection in the Endeca UI Design Pattern Library. In this post, I’ll provide a bit of background to these additions and outline the ways in which we had to extend the conceptual framework of the library itself to accommodate this new material.

First, the patterns themselves. This new set consists of three patterns focused on spatial information visualisation and one on design principles for analytic applications:

1. Point Location Maps help users perceive spatial patterns in record location, identify specific records for further investigation or action, and explore relationships between particular facets and record locations within a broader spatial area.  For example:
   • Where are our top performing agents located?
2. Region Maps help users perceive spatial patterns in record distribution, understand how those patterns relate to pre-defined boundaries and regions, and explore relationships between particular facets and aggregate distributions across a broader spatial area. For example:
   • In which states are our sales above average?
3. Heat Maps help users perceive areas of greater or lesser record density, examine degrees of variation based on spatial factors, and explore relationships between particular facets and aggregate patterns of density across a broader spatial area. For example:
   • Where are the most intense areas of traffic congestion within the city?
4. Analytics Applications summarize important metrics and trends and aggregate key quantitative and qualitative information sources, providing visibility and information scent through faceted visualizations (e.g., dynamic charts, graphs, etc.), metrics tables, refinements and other analytic summaries.

But one of the most interesting issues behind these patterns is the changes required to the library itself in order to make this new material navigable and searchable. As you may have noticed, patterns are currently categorised using three primary facets: Industry, Topic, and Usage. It is the latter two I wish to focus on in this post.

Let’s start with Topics. This facet originally had six values:

1. Faceted Navigation
2. Promotional Spotlighting
3. Results Display
4. Results manipulation
5. Search
6. Signposting

This set of facets reflected the original scope of the library and in some ways also surfaced our own world-view of the conceptual landscape of information search and discovery. But now, with an increasing focus on analytics and Agile BI, we’ve extended that framework to accommodate the new topics of Spatial Visualization and Faceted Analytics.

Now let’s consider Usage. The aim of this facet is to represent the purposes to which we expect each pattern to be applied, and in that respect it closely mirrors the Modes of Interaction we discussed in an earlier post. Originally, this facet was assigned the following eight values:

1. Analyzing
2. Comparing
3. Evaluating
4. Exploring
5. Locating
6. Refining
7. Sharing
8. Validating

However, defining such a taxonomy (even one as modest as this) is not an exact science – categorisation schemes reflect a subjective view of the world, and in that respect, who is to say this view is any more authentic or valuable than anyone else’s? Even well-known academic frameworks such as the search activities defined by Gary Machinioni reflect some degree of subjectivity in the interpretation of the research evidence:

Three Kinds of Search Activities

 

That said, some immediate shortcomings of the original taxonomy were becoming apparent. Well-chosen facets should induce high entropy in the result set (e.g. through being consistent, orthogonal, exhaustively applied, etc.), and in that respect two of the values just didn’t seem appropriate:

• “Refining” was too low-level and task-focused, lacking the goal-directed nature of the others
• “Sharing” was too generic (applicable to almost any discovery scenario), and seemed to apply to a different level of the discovery process

This conclusion was coupled with the observation that our own research had identified usages that didn’t fall into any of the above categories, such as:

• “Monitoring”, i.e. maintaining awareness of the status of an item or data set for purposes of management or control, e.g. “I need to monitor failing customers/dealers so I can prompt my Account Reps to fix the problems”
• “Synthesizing”, i.e. generating or communicating insight by integrating diverse inputs to create a novel artefact or composite view, e.g. “I need to prepare a weekly report for my boss of how things are going”

So applying these modifications, along with the extensions to the Topics, we see the complete set of facets that we see on the UIDPL site today.

As mentioned above, no categorisation scheme is ever perfect, and they all to some degree reflect a subjective world view. And no doubt these facets will evolve further as our collective understanding of human information-seeking behaviour develops.  Instead, the real measure of their value is the extent to which they facilitate practical goals and tasks. So if you find them meaningful and valuable in using the pattern library, then they have fulfilled their purpose. If you think they could be improved, we’d love to know.

Related Posts:

1. The Dimensions of Search User Experience
2. User Interface Design Patterns for Search & Information Discovery
3. The Changing Face of Search
4. Search Solutions 2010: Highlights & Reflections
5. Search at the Guardian Newspaper