Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watnot!watmath!clyde!rutgers!mit-eddie!uw-beaver!tektronix!sequent!mntgfx!franka
From: franka@mntgfx.UUCP
Newsgroups: comp.ai
Subject: Re: analysis of unknown data
Message-ID: <564@franka.mntgfx.MENTOR.COM>
Date: Wed, 11-Mar-87 13:38:32 EST
Article-I.D.: franka.564
Posted: Wed Mar 11 13:38:32 1987
Date-Received: Fri, 13-Mar-87 04:12:26 EST
References: <5681@mimsy.UUCP>
Reply-To: franka@mntgfx.UUCP (Frank A. Adrian)
Organization: Mentor Graphics, Beaverton, OR
Lines: 29
Keywords: statistics; pattern recognition; signal processing; content analysis

In article <5681@mimsy.UUCP> dave@mimsy.UUCP (Dave Stoffel) writes:
>
>
>    What systematic methods and techniques would you apply to the
>    following problem?
>
>    Determine the representation, organization, and content of a
>    "file" containing up to 156MB.  There are no assumptions.  The
>methods and techniques applied must be automated (if not fully
>automatic) and applicable to an unlimited supply of "files".

Actually, there are several ways to approach this problem.  It is a statement
of finding out what is happenning inside a classical "black box".  You can
start by monitoring all requests and replies from the file, searching for
patterns based on location of access and length of access.  You can examine
the bytes returning from the device to try to detect patterns.  You can use
a traffic analysis approach by find out what types of programs access this
file at which times for a given purpose.  You can go ask the NSA, CIA, and
other intellegence agencies what they do when they try to crack a black box
(though I doubt that they'd tell you :-).  Finally, most boxes are not com-
pletely black.  In general, you can tell information by the location, size,
etc. of a box.  But unless the box is completely isolated (in which case, why
are you all that interested in what it does?) you can always get some infor-
mation, upon which you can make your own assumptions, can try experiments,
and finally uncover the nature of an object.  You might also try any good
text on experimental methods to point you in the right direction.

Frank Adrian
Mentor Graphics, Inc.