Re: RDS web-page search
Lewis Townsend writes:
> Well, I guess the main thing I was wondering is if it would
> be too hard to expand the capabilities of the search engine
> on the RDS website.
Each time you run it, the search program reads the HTML on the
Recent User Contributions and Archive pages.
It strips out most, but not all, of the "noise", such as formatting
information, font etc. For each entry, it counts the number of
matches against the words that you type. There is no
"and" or "or" logic implemented. Quotes are not
handled properly either, so you can't search for phrases, just words.
I suppose "a b c" will search for a single word: "a b c" that must
appear on one-line of HTML - not that useful.
Case is not significant. If a word contains less than
3 characters, it must match a full word, not a substring.
The score for an entry is the sum of the square roots
of the individual word-counts. e.g. searching
for Lewis Townsend, if "lewis"
occurs 4 times in an entry, it only counts as 2 points.
An entry with both "lewis" and "townsend" will
score 1+1=2. An entry with
"lewis" twice, but no "townsend" will only
score 1.414 (root of 2).
Thus entries that match multiple distinct words
will tend to be favored over entries that match the same
word many times.
You will often get a huge number of matches,
but we sort the results by score, and feed them out to you
15 at a time, so they start to appear quickly.
Usually only the top few items are relevant, but it doesn't
hurt to get more (you can always press Stop on your browser).
We've had requests to add a check box to select
DOS or Windows or Linux or any combination,
but I didn't feel it was that useful, since many of the
programs labelled as DOS will actually work on 2 or
3 platforms, not just DOS, plus it's not hard to visually
filter the colors in the search results.
Regards,
Rob Craig
Rapid Deployment Software
http://www.RapidEuphoria.com
|
Not Categorized, Please Help
|
|
