Usenet Search Engine Tutorial

[D Finnigan <dog_cow@macgui.com>]

Here is a tutorial on using the enhanced Usenet search engine at Mac GUI.
Let's say that you are interested in doing some research about various Apple
II accelerators.

You start off by entering a single keyword:

transwarp

This will of course match all articles that contain the word transwarp, and
it will bring a lot of results. To reduce duplicate results, quoted text
will never match any search terms. Since there are so many results, you
decide to narrow your search to articles that talk about upgrading the
transwarp. You search for:

transwarp upgrades

This will match articles which contain both words transwarp and upgrades.
However, similar words, such as upgradeable, upgrading, upgrade, etc, will
also match. Certain words that are very common, such as 'a', 'and', 'the',
are stopwords, and they are ignored in keyword searches.

At this point, you want to match articles that very precisely mention
upgrading the Transwarp, so you enter an exact phrase search:

"upgrade the transwarp"

This search will only return articles with these exact three words in
sequence. As opposed to an ordinary keyword search, alternative word forms
will not match, so an article containing the phrase "upgrading the
transwarp" would not match this query. However, stopwords do match. You can
use these two characteristics to your advantage, for if you want to match a
single word exactly, you can enter it as a one-word exact phrase, as in this
example:

"upgrade the transwarp" "configuration"

Of course, this means that it is possible to combine several exact phrase
searches in a single query, and only articles which contain all of those
phrases will match.

Phrase searching also allows for any number of wild cards within the phrase.
Let's say that you want to search for some articles containing a phrase
including the word cache and the name zip-chip. You enter two wild cards to
allow for any two words in between:

"cache * * zip-chip"

This query will match articles that contain the phrase "cache for a Zip
chip" or "cache on the Zip-Chip" or any other permutation. Also notice that
the hyphen is dropped; it will match both Zip Chip and Zip-chip.

Now you want to find articles that will tell you whether the Transwarp III
really existed or not, so you enter:

"transwarp iii" existence

Now only articles which contain the exact phrase "transwarp iii" and the
word existence (or a variation thereof, such as existed) will appear.

It is also possible to exclude words. Doing so will tell the search engine
to only return articles that do not contain one or more words. Simply type a
minus sign in front of a word to exclude it from the results. Excluding a
word will also exclude all of its forms and tenses. Although finding
articles that don't contain the word 'the' may be amusing, directives to
exclude this and other stopwords will be ignored.

Now you want to find articles that only mention the Zip Chip, and not the
Transwarp, so you search for:

zip chip -transwarp

And sure enough, all of the articles that result contain the words zip and
chip, but not transwarp. Articles that contain variations on transwarp, such
as transwarps or transwarping are also excluded from the search results.
Special care should be taken when excluding words and searching for an exact
phrase. Here is an exact phrase search where one of the words in the phrase
is excluded:

"transwarp gs" -transwarp accelerator

Can you guess what the result will be? No results will match because this is
an impossible query! Similarly, it is illegal to have a search query
consisting of nothing but excluded words. A current limitation also does not
permit an exact phrase search combined with only excluding words, such as:

"zip chip" -transwarp

This query would only search for the exact phrase; the directive to exclude
the word transwarp would be completely ignored. However, this search is
entirely legal:

zip chip "pc transporter" -6502

This query will return articles which don't contain the word 6502, but
contain the exact phrase "pc transporter" and the two words zip and chip. It
should also be mentioned that the order of the words in a search query
(except for inside of an exact phrase, of course) does not matter.

In certain situations, in order to return more relevant results, the search
engine will ignore certain words of your query. If you absolutely insist on
matching these words in the search results, type a plus sign in front of the
word, as in:

"transwarp" "Nucleus" +big screen

All of the searches so far did not match words in the subject of any
article. If it appeared to do so, it was only by coincidence. To explicitly
match words in the subject, use the subject: operator, like so:

subject:mhz

This will return articles with the word MHz in the subject. Only one word
may be used per subject: operator, and the word must follow immediately
after the colon. Therefore, to match '9 MHz' in the subject, enter:

subject:mhz subject:9

As with everything else, the subject operator may be combined with any exact
phrases or keywords to match fewer articles. Just like regular keywords,
similar word forms will also match. There is no way to specify words to
exclude from the subject, nor to specify an exact phrase match.

It is possible to return results from matching authors by using the author:
operator. As with subject:, only a single word should be used with each
author: operator.

speeddemon author:parik

To restrict articles to those only from one or more groups, use one or more
group: operators. You may specify a wild card using the asterisk to match
multiple groups in a hierarchy. You may use more than one group: operator,
and results from any of the matching groups will be shown.

titan accelerator group:comp.sys.apple2

Ordinarily, all search results are ranked by descending order of relevance.
The relevance is computed using word frequency scoring. Words appearing in
the subject are ranked higher than those appearing in solely in the body.
Alternatively, it is also possible to rank search results in descending
order of date. In this latter case, no word frequency scoring is performed.

-- 
]DF$
Apple II Book: http://macgui.com/newa2guide/
Mac GUI Vault: http://macgui.com/vault/
Usenet Archive: http://macgui.com/usenet/