- Time Slices
Suppose you visited the archaeology museum in
Heraklion on
the isle of Crete and were looking at a particularly interesting piece of early
Cycladic pottery. Afterwards, you wanted to
know what did the contemporary pottery in Japan look like? Or in North America?
Where were there telegraphs in 1878? What did population density look
like in 14 BCE? Create an animation of the boundaries of the Roman Empire
from the time of Augustus to the time of Justinian. And throw in the other
major political contemporaries (oh, but who were the political
contemporaries of the Roman Empire? Crate a list.). Let the user choose the
color scheme.
This system will design and implement
tools so that the user can explore history in as flexible a manner as possible from
the comforts of a laptop.
- Never Throw It Away
How much of your hard drive would be used up if you didn't delete any of
your email for a year (both sent and received)? How many bytes to store the
email addresses of everyone you've ever exchanged email with? Why delete
todo items? After all, don't you think 10 years down the road you might want
to be reminded that you called the exterminator on 3 Feb 1993? What was your address
on 3 Feb 1993? You'll need it to fill out that application for a security clearence. The fact is
that hard drives are cheap enough that there is no point in deleting
anything.
One reason for not keeping all this information around is that there
aren't good tools for managing and navigating hordes of trivial dribs and
drabs of information. Perhaps a more important question is what would you do
with all these bits if you did keep them around?
This system will collect iformation for your calendar, email, to-do lists, etc. and allow you to easily recall information about your past, discover patterns in your past behavior, and explore the minutiae of your life.
- CodeSearch
This project will result in a "google"
for the source in open-source projects.
There is now a gigantic amount of open-source code available.
The goal of this project is twofold: (1) Search for and identify
similarities in code; (2) Search for an identify code that is
structurally similar to a search expression. Thus, suppose you
are writing code to scan a string and look for blanks. You KNOW
this has been done before - but where to find it? One way is
to download all open-source projects one by one from SourceForge
and grep on comments. A better way is to create a "structural search"
tool:
- Enter the search term:
boolean searchForBlanks (char *)
{
/* Look for blanks in a string */
for (int i=0; i
- The search should find actual functions that look like:
int func_String_bl_detect (char[] inputStr, int strLen)
{
/* Blank detection */
for (k=0; k
- How does it do that? By a combination of word search (look for the
word "blank" within comments) and structural comparison using parsing.
As part of this project, you will scour through tons of open-source
code (automatically, using your tool, of course) and collect
statistics on usage patterns—idea proposed by R. Simha.
- Memoir: Context-Archival
This is another
linux-kernel type of project, but also involves modifying a
browser like Firefox. The goal is to store context information
in the Unix file system, perhaps using meta I-nodes or something
like that. The thinking is that, archival systems today merely
store files without context information. This will make it very
difficult for future historians to understand the purpose and
context of files. For example, a historian examining the White
House electronic file archive 20 years from now will have a
terabytes of files to try and make sense of.
This project will explore the idea of storing context information
so that one can understand where files came from, what other
files they relate to, what happened at the time of
creation/modification etc. For example, if future
historians examined the file /users/home/president/WarJustification.pdf,
they would like to know where it came from (from email? from whom?
downloaded from a website? which one?) and what other files were
being used in context at that time (saddam.gif?).
In this project, you will modify the Linux kernel and a browser
to make this combination work, and will examine the overhead
involved (how does this slow down file accesses?)—idea proposed by R. Simha.