subscribe
PLAN: Personal Life Activity Notation
Since people naturally use narrative to convey past events to others, we thought that allowing Poyozo to represent captured user activity data in textual narrative form would allow this data to be more immediately meaningful and evocative. This led us devise PLAN, a notation which, like the proposed activitystrea.ms format is used to express sequences of sensed activities, but which is designed for human-consumption as well– specifically, it is designed to be easy for people to read, edit, and author. Reducing the barrier to authoring and editing captured data is important to the system because it enables users to effectively engage in a dialog with the system, fixing erroneous sensor data, supplementing activity logs with new data (sensor-unobservable) data, and deleting undesired records.
PLANs syntactically look like meeting agendas: each PLAN is an ordered sequence of consecutive events under a common “episode” (short period of time, analogous to a meeting), where each event consists of a timestamp, optional subject, action (predicate), object, and optional duration. When the subject is omitted, the user’s own self is implied; this allows entries to remain succinct and easy to read. Example PLAN sequences and full spec coming soon.
Raw captured activity logs generated by sensors often are extremely detailed and voluminous; the sheer quantity of such entries often makes reading PLANs tedious and overwhelming. For example, Poyozo captures every action performed by the user as s/he navigates the web, including querying search engines, switching tabs, and intermediate pages viewed while seeking information.
Thus instead of dumping raw activity logs as PLANs, Poyozo generates PLANs as responses to queries which can restrict and filter which events are selected in various ways. Queries can select events by subject (user performing the action), action type, or object that the action is being performed upon. Furthermore, queries can be set to include only events that take more than a certain duration, or which exceed a combined uniqueness-importance measure analogous to TF-IDF for document retrieval. For filters with thresholds, Poyozo provides a graphical slider which can be used to interactively expand or restrict returned results.
created in collaboration with Max Van Kleek
Comment
-
about me
I am investigating interfaces, interaction techniques, and visualization methods that help people understand their lives through exploring their own personal information
Check out my current projects - atomate, eyebrowse and poyozo
email: brennanmoore@gmail.com
twitter: @zamiang
more info -
tags
-
thoughts
-
topics
