Designing an information-finding strategy
Your chosen strategy should start with a decision: what mode are you in?
Hunting or Harvesting?
- Hunting for information:
- You know exactly what you are looking for
- You have some idea what it is called
- You have some idea where it might be
- Experienced searchers can find their prey - if it is there!
- Harvesting information:
- You don't know exactly what you are looking for
- You have little idea where it might be
- You might recognise it when you see it
These two 'modes' of searching are SO DIFFERENT as to merit completely different
strategies and tools
Once you have decided what mode you are in, you need to think about your target.
How can you describe it? What do you know about that information that can let you
guess where it might be? Is it likely to be in a certain country, or published by a
certain type of organisation? (e.g if you know it is a UK academic site then it is likely
to have .ac.uk in the URL)
Then you need to think about some words which would likely appear in the title and/or
text of that page. Make a note of these: you should try to use tems which you think
will be as unique as possible for that resource. And if you can't think of anything
unique, use at least 6 search terms, in order of their likely importance.
Don't expect Web publishers to have appropriately described or classified the
information!
This is where the Web really differs from an online information service. If you
searched Dialog for market research about the blood transfusion industry, you might expect
the information to be 'tagged' as market research about the blood transfusion
industry. But on the Web the best way is often to search not by a description
of your resource but rather, by using words which would actually occur in the document you
seek. So in this example, market research about the blood transfusion industry would
mention certain phrases: "blood processing equipment" "blood
transfusion" "apheresis" "leukofiltration", for example.
|
|
Constructing a query: Hints and tips
When hunting:
- Start at a large search engine: Altavista, Northern Light or Hotbot or a metasearch
engine
- Your best bet to find a needle in a haystack is to use a magnet - if you are going to
get 11,253 search results then make sure that the result you need is in the top ten!
- To make the most of the statistical search engines, don't use Boolean queries - they can
skew the all - important relevancy ranking.
- Instead of Booleans try to use specific phrase searching, if you are fairly
certain that your result would include that phrase. For example: "
castles in Pembroke, Wales" instead of castles AND Pembroke AND Wales OR,
try using a number of search terms which would be usually found in your target page and
simply list them in order of importance, and search for ALL of these terms (if asked to
select ANY or ALL)
- If you are sure about which terms you want to include and exclude, use Search Engine
Math (also known as 'standard syntax')
- For precision, try field-restricted searching, where you search for the term in a title,
link or URL (e.g. title:searchterm url:searchterm and link:searchterm in Altavista,
ti:searchterm u:searchterm in Yahoo)
When harvesting:
- Start at a good quality directory, such as Yahoo,
About.com, (www.about.com) the
Google Web Directory
(directory.google.com)
- Try to think broadly about the general subject of your search. Who might publish
the information you need? What kind of Web site might it be found at? Then look for
that wider resource. This has the benefit of giving you not only what you need but other,
well organised documents around the same topic.
- Avoid Booleans especially when in harvesting mode. Instead, think of three or more
descriptive words or phrases and search for ANY of them (in standard syntax this means do
NOT use + or -)
|
