Next Step in Selecting Topics and Group: Dynamic Personal Information Spaces Project Topics

The specifics of your project will evolve over the quarter but it is important to get started as soon as possible. A quarter is a really short time period. As is common in research, you often need to commit to a project idea before it is fully elaborated. Groups need to start to form as early as possible. Group membership and project topic selection absolutely must be completed by week four. Establishing groups and decidiing on a topic before week four is much better.

All projects must explore some aspect of the notion of a dynamic graphical personal information space as discussed in the NSF proposal. One way of thinking about this space is as a separate visual space composed of rule-governed entities that are derived from and linked to existing information and applications. In our research group meetings we sometimes characterize the space as one that abstracts and removes the messiness of the web and existing applications to better enable integration and personalization of information. In addition, we argue that this space has potential to provide a new form of interoperability between applications. Normally one thinks of needing to change applications to make them interoperable. There are, of course, many factors that make that difficult. They range from access to source code to commercial competitiveness factors that often result in walled gardens. The fundamental idea is to abstract aspects of existing information and reify them in a separate linked space as entities that operate according to multiple composable and tailorable context-sensitive rule sets that determine how they look, behave, and interact. This separate space can not only provide mechanisms for integration and personalization of information but also a new form of interoperability at the level of these new information entities rather than necessitating changes to applications. That is to say some of the benefits of application interoperablity can result from designing rules in the parallel linked information space that operate in ways that help coordinate information between applications.

As part of this step in selecting a topic and group, we will first discuss together and then break into groups to further discuss and brainstorm possible project. Here are possible project topics to prime brainstorming:

• Representation and Behavior of Information Entities
  Select a type of information entity (e.g., PDFs, browser tabs, emails, calendar entries, to-do lists, code files, videos, notes, presentation slides, search results from Google Scholar or other searches, etc.) and explore effective ways of representing these entities and how they should behave in a multiscale dynamic personal information space. You might develop representations in terms of different zoom scale or how they behave in ways sensitive to activity, task, or other semantic context. For example, a PDF might be represented as a graphic icon that showed the first page of the PDF so one could see the title and author with perhaps a montage of images from the PDF arrayed below and how it might chage as a result of semantic zoom (e.g., at a sufficiently zoomed-in level, or perhaps on hover, the iconic representation might change to show the main headings of the paper.)
• Organization and Behavior of Groups of Information Entities
  Explore grouping behaviors of information entities. For example, placing an entity close to a group of entities might add it to the group. How might that be represented and how should group members make space for a new element being added? For example, as one positioned an element other elements might move out of the way to make room according to a force layout behavior. One might constrain the size of a group and elements might zoom down as new elements are added. An interesting semantic behavior would be as the group is zoomed down, group prototypical elements might not scale or even get larger as other elements scaled down to better represent the group.
• Activity Histories
  Explore ways to represent activity histories. For example, this might be the history of adding tabs in a browser. How might tabs be represented in a separate personal information space? How could activity within a tab or between tabs change the representations in ways that are useful? Might multiple movements between a set of specific tabs impact their representations or organization? Any activity history (e.g., editing a file, or reading an article) would be interesting to represent and explore useful behaviors for those representations. Consider new ways of representing a collection of files unconstrained by the usual hierarchical organization. For example, similar to tabs access to files could generate representations in a personal information space. Especially challenging (and interesting) is cross-application activity. Currently history within an application is captured but real activity involves multiple applications and that is not explicitly captured or represented. Be interesting to explore representation of this activity.
• Lenses
  Explore creating lenses through which entities are viewed. For example, imagine file entities being re-represented when you position a lens over them that makes use of specific metadata such as file creation or last access date (e.g., they might move to an appropriate place along a timeline). Because we want the rules of the information space to be composable, it would be especially interesting to explore composition of lenses (e.g., spatially combining a lens that grouped by file type with one that ordered by creation date by placing the file-type lens on top of the creation-date lens). Lenses provide a visual metaphor for querying metadata. One can also think of lenses not just as presenting temporary views of entities within them but as enabling different ways of interacting with the elements (e.g, reaching through a lens one could interact with some entities to change the way they depict themselves such as the semantic level of their representation or a lens might alter the layout of selected entities within it.) Of course, these lenses could also be composed. Another way of thinking of lenses is as a visual metaphor for unix pipes.
• Zoomable Multiscale Personal Information Space Infrastructure
  For groups who would like to focus on infrastructure, here are some suggestions for projects:
  • Explore creating a browser-based zoomable multiscale information space that support some of the functions of Pad++ . Here is a short and very poor quality PadCad video we created early on to demonstrate some of Pad++ features. It is more than you likely want but here is the Pad++ Reference Manual.
  • Expore exploiting Webstrates based information space to provide collaboration and cross-device interaction. Webstrates is being developed by our collaborators at Aarhus University.
  • A couple of grad students have made considerable progress on getting Dynapad, the last version of Pad++, running. Dynapad was a Scheme wrapping of the Pad++ C++ rendering engine. It would be fabulous to have a group finish this project. The group would need strong C++ coding skills and some knowledge of Scheme. The version of Scheme used for Dynapad was PLT Scheme which has now evolved into Racket. Racket is desribe as a "language-oriented programming language" and is suppported by an excellent group of progrmaming language researchers.