Research ideas

| -Uncategorized
Blog visualizer

It is useful to see how blog entries and other chronologically
arranged items like e-mail cluster around certain topics over a period
of time.

Your system should produce a graphical view of an RSS or Atom XML feed
containing blog entries (or any other chronological data), optionally
categorized (possibly multiple categories). Categories will be
represented as rows in the table and time will be represented in
columns. See http://kmdi.utoronto.ca/RMB/papers/p11.pdf and the
Lifestreams project for ideas.

The system can be split into two parts. One pre-processes the XML
feed, the list of categories, and the time definition, producing
another XML document containing the data in table format. The other
part visualizes the data from the first part. Alternatively, the
entire system could be coded as one component.

The system should support:

– autodetection of categories using the attribute
– categories defined as keyword searches on the entry title and body
– a list of pre-defined categories

A nice plus would be the use of ontologies to map categories onto
other categories. =)

The system can be a standalone application or a web-based application.

Wearable personal information manager

An easy-to-use personal information manager has not yet been developed
for wearable computing use. Wearable computers have limited input and
output capabilities. Notably, mouse use is very limited. Speech
recognition can make accessing personal data such as tasks,
appointments and notes easier and more intuitive, and simplified
interfaces allow users to manipulate personal information even while
walking around.

The system should allow the user to manage tasks, appointments, and
notes. Other functionality can be added as desired.

Your research will integrate a limited-vocabulary speech recognition
engine such as CMU Sphinx with a basic personal information manager.
You may develop a personal information manager from scratch or adapt
existing open-source software. (I maintain a PIM, so I can point you
in the right direction.)

The system should be a standalone application, as network connectivity
may not always be available and the web browser interface is too
limited.

The system can be prototyped on desktop computers and targeted either
for heads-up-display+audio (320×240 b/w, synthesized speech) output or
pure audio output.

Random information manager

Develop a random information manager that allows users to store
unstructured text, browse, edit, and assign multiple categories to
them, and produce RSS or Atom feeds based on ‘slices’ of this data
determined by keyword search, chronological bounding, or category
filtering. The system should allow the user to explicitly exclude
certain items from all automatic feeds (private entries) as well as
from particular feeds.

The system can be a standalone application or a web-based application.

You can comment with Disqus or you can e-mail me at sacha@sachachua.com.