 |
|
|
Yes. It's an interesting ability to decrease the weighting of certain words, quite different from setting Stopwords which get removed from the search entirely. By setting Less Important
words, you will still get the same number of search results because all
of the words are still being searched. The concordance (text excerpts)
will still highlight all of the words that the user typed. But the order
of search results will emphasize the other words in the search,
bringing them up closer to the top. And if the user only searches for
Less Important words, the results will not be affected.
When would you want to try setting Less Important words? The statistical
ranking algorithms of PicoSearch will naturally emphasize the more
distinctive and rare words in a search. So most searches should return
the most relevant documents first or nearly first, and you'll never have
to consider this or other weighting factors.
But because PicoSearch recognizes that not all websites are created
equal, we provide many tools for customizing your search engine. Factors
such as unusual uniformity in document text, or very small documents
(such as just searching titles), can minimize statistical differences
and call for additional search strategies. If you notice common searches
not performing as you'd like, we suggest that you read our FAQ on influencing the search. Some options are more global in scope than Less Important words.
In general, if your search engine is performing fine except for a
certain word which seems to overwhelm the results whenever it is typed,
that could be a good candidate for a Less Important word. For example,
on a hair care site, maybe so many of the searchable words are equally
common that the word "hair" doesn't de-emphasize as much as you'd like
when the user types [hair product] or [hair brush], although [hair gel]
might be fine with the rarer word "gel". Then it certainly wouldn't
hurt to make "hair" a less important word.
Less Important words do allow wildcards. ? matches one optional
character, * matches any number of optional characters. So a pattern of
auto? would match "auto" or "autos" but not "automobile", whereas auto*
would match all of these plus "automation" and more.
Multiple word patterns are allowed for increased precision to match
consecutive terms in the search. No quotes are needed, just spaces
between the words. So for example a pattern of auto* race? would match
"automobile races" but not "automated dog races".
Try it, and feel free to ask us if you have any questions.
|
|
 |
|
|
