Search strategy: Lightning Hypertext of Disease

Jules Berman jjberman at ix.netcom.com
Sat Sep 14 15:52:13 EST 1996


SEARCHING THE LIGHTNING HYPERTEXT OF DISEASE SEARCH ENGINE

Our web site is:       http://www.pathinfo.com/

Since we opened our free pathology search engine, we have had
thousands of inquiries.  Each inquiry is logged onto the server
for our review, and we have noticed a few common mistakes that
lead to nonproductive search queries.

We are listing them here so that the users of bionet.toxicology can
avoid any frustration.

1. DON'T USE BRITISH SPELLING

The Lightning Hypertext is written in American English, so it's

hemochromatosis, not haemochromatosis
nevus, not naevus
esophagus, not oesophagus
etc.


2. DON'T USE SINGLE WORD QUERIES CONTAINING A COMMON 5-LETTER
LEADING STRING.

The free version of our hypertext performs leading string searches
on the first five letters of each search word entered.

If you enter `neurology', you will get every information packet that
contains the prefix `neuro', including `neurotic, neurology,
neurosurgery.'

If you enter `prostatic', you'll also get any information packets
that includes `prostitute', as well as packets containing prostate-
related terms.

The free version of the search engine limits the user to
30 `hits' per query.  Therefore, a legitimate query word might
yield only irrelevant cards if the query term has a common
leading string. For instance, a search on the single word
`neurothekeoma' would yield hundreds of information packets
for the many words beginning with `neuro'.  It is unlikely that
`neurothekeoma' would appear within the first 30 information
packets produced as output by the search engine.  Entering the same
word again will not produce the next 30 `neuro' packets.  It will
only produce the same 30 information packets you have already seen.

SEARCH AMBIGUITY CAN BE OVERCOME BY CLEVER USAGE OF ADDITIONAL
SEARCH WORDS WITHIN THE QUERY.

So if you are using a word with an ambiguous 5-letter leading
string (such as `neurothekeoma'), you need to add a second word
related to the first word that will limit the search, such as
`neurothekeoma myxoma'.

For many words with common prefixes, such as neurothekeoma,
we have adjusted our search software so that you can just enter
`neuro+thekeoma' and you'll get your search results.  Try it.


3.  DON'T USE UNLIKELY QUERY STRINGS

The Lightning Hypertext of Disease is a powerful search engine,
but it is not magical.  It consists of about 29,000 information screens
related to pathology, and it collects the screens that contain all
the leading strings of all the query words entered.

So if you enter `informatics university public domain' because you
happen to be interested in the field of informatics, and want
to know what other universities are doing in this field, and only
want public domain text, you are likely to be disappointed by
the results.

We suggest starting with a general search, and then narrowing your
search based on the output from the general search.  You might
start with `informatics'.  At the top of the output page, the number
of collected information packets will be listed.  If the number
is high (over 100), you might think of narrowing your search.
Based on browsing through the information packets produced, you might
want to branch off on some term present in one of the information
packets.

Remember, all of the words of the Lightning Hypertext are searchable,
except for the most common words, such as `and, when, which, for, etc.'
So almost anyting you see in any information packet will be valid
words for a new query.

4. DON'T USE THE LIGHTNING HYPERTEXT OF DISEASE IF YOU ARE NOT A
HEALTH PROFESSIONAL.

Our product is not designed to help laymen, and you will only be
disappointed by the output.

5. IF YOU DON'T KNOW WHERE TO START, TRY SOME OF THESE SEARCHES:

legal
oncogene
mumps (the language, not the disease)
preneoplasia
statistics epidemiology
pain
research
liver transplant
chromosome abnormality
administration
prevalence
mutation
toxicology
coding
snomed
proficiency

Good luck, and please send us feedback.

Jules Berman, President, Pathology Informatics, Inc.





More information about the Toxicol mailing list