How to use the Keyword Search
Natural Language Searching
This is similar to the search syntax used by many popular Internet search engines.
A natural language search request is any combination of words, phrases, or
sentences. After a natural language search, Keyword Search sorts retrieved documents
by their relevance to your search request. Weighting of retrieved documents
takes into account: the number of documents each word in your search request
appears in (the more documents a word appears in, the less useful it is in
distinguishing relevant from irrelevant documents); the number of times each
word in the request appears in the documents; and the density of hits in each
document. Noise words and search connectors like NOT and OR are ignored.
A plus in front of a word or phrase indicates that the word or phrase must
be present, and a minus in front of a word or phrase indicates that the word
may not be present. Example:
+"first class mail" +postage -meter
This would find any document that contains "postage" and "first class mail" but not "meter".
Overview of Boolean Search Syntax
NOTE: To use the following boolean syntax you MUST select Search Type="Boolean" in the form above.
A boolean search request consists of a group of words
or phrases linked by connectors such as and
and or that indicate the relationship between them.
|Both words must be present
|Either word can be present
|Apple must occur within 5 words of pear
|Apple must not occur within 5 words of pear
apple and not
|Only apple must be present
|The field name must contain smith
If you use more than one connector, you should use parentheses to indicate
precisely what you want to search for. For example, apple and pear or orange
juice could mean (apple and pear) or orange, or it could mean apple
and (pear or orange).
Noise words, such as if and the, are ignored in searches.
Search terms may include the following special characters:
|Matches any single character. Example: appl? matches apply or apple.
|Matches any number of characters. Example: appl* matches application
|Stemming. Example: apply~ matches apply, applies, applied.
|Fuzzy search. Example: ba%nana matches banana, bananna.
|Phonic search. Example: #smith matches smith, smythe.
|Synonym search. Example: fast& matches quick.
|Variable term weighting. Example: apple:4 w/5 pear:1
Words and Phrases
You do not need to use any special punctuation or commands to search for a
phrase. Simply enter the phrase the way it ordinarily appears. You can use a
phrase anywhere in a search request. Example:
apple w/5 fruit salad
If a phrase contains a noise word, Keyword Search will skip over the noise word
when searching for it. For example, a search for statue of liberty would
retrieve any document containing the word statue, any intervening word,
and the word liberty.
Punctuation inside of a search word is treated as a space. Thus, can't
would be treated as a phrase consisting of two words: can and t. 1843(c)(8)(ii)
would become 1843 c 8 ii (four words).
Wildcards (* and ?)
A search word can contain the wildcard characters * and ?. A ?
in a word matches any single character, and a * matches any number of
characters. The wildcard characters can be in any position in a word. For
appl* would match apple, application, etc.
*cipl* would match principle, participle, etc.
appl? would match apply and apple but not apples.
ap*ed would match applied, approved, etc.
Use of the * wildcard character near the beginning of a word will slow
Use the AND connector in a search request to connect two expressions, both of
which must be found in any document retrieved. For example:
apple pie and poached pear would retrieve any document that contained both phrases.
(apple or banana) and (pear w/5 grape) would retrieve any document
that (1) contained either apple OR banana, AND
(2) contained pear within 5 words of grape.
Use the OR connector in a search request to connect two expressions, at least
one of which must be found in any document retrieved. For example, apple pie
or poached pear would retrieve any document that contained apple pie,
poached pear, or both.
Use the W/N connector in a search request to specify that one word or phrase
must occur within N words of the other. For example, apple w/5 pear would
retrieve any document that contained apple within 5 words of pear.
The following are examples of search requests using W/N:
(apple or pear) w/5 banana
(apple w/5 banana) w/10 pear
(apple and banana) w/10 pear
Some types of complex expressions using the W/N connector will produce
ambiguous results and should not be used. The following are examples of
ambiguous search requests:
(apple and banana) w/10 (pear and grape)
(apple w/10 banana) w/10 (pear and grape)
In general, at least one of the two expressions connected by W/N must be a
single word or phrase or a group of words and phrases connected by OR. Example:
(apple and banana) w/10 (pear or grape)
(apple and banana) w/10 orange tree
Keyword Search uses two built in search words to mark the beginning and end of a
file: xfirstword and xlastword. The terms are useful if you want
to limit a search to the beginning or end of a file. For example, apple w/10
xlastword would search for apple within 10 words of the end of a
NOT and NOT W/N
Use NOT in front of any search expression to reverse its meaning. This allows
you to exclude documents from a search. Example:
apple sauce and not pear
NOT standing alone can be the start of a search request. For example, not
pear would retrieve all documents that did not contain pear.
If NOT is not the first connector in a request, you need to use either AND or
OR with NOT:
apple or not pear
not (apple w/5 pear)
The NOT W/ ("not within") operator allows you to search for a word
or phrase not in association with another word or phrase. Example:
apple not w/20 pear
Unlike the W/ operator, NOT W/ is not symmetrical. That is, apple not w/20
pear is not the same as pear not w/20 apple. In the apple not w/20 pear
request, Keyword Search searches for apple and excludes cases where apple
is too close to pear. In the pear not w/20 apple request, Keyword Search
searches for pear and excludes cases where pear is too close to apple.
Numeric Range Searching
A numeric range search is a search for any numbers that fall within a range.
To add a numeric range component to a search request, enter the upper and lower
bounds of the search separated by ~~ like this:
apple w/5 12~~17
This request would find any document containing apple within 5 words
of a number between 12 and 17.
Numeric range searches only work with positive integers. A numeric range
search includes the upper and lower bounds (so 12 and 17 would be
retrieved in the above example).
For purposes of numeric range searching, decimal points and commas are
treated as spaces and minus signs are ignored. For example, -123,456.78
would be interpreted as: 123 456 78 (three numbers).
Stemming extends a search to cover grammatical variations on a word. For
example, a search for fish would also find fishing. A search for applied
would also find applying, applies, and apply.
There are two ways to add stemming to your searches:
- Check the Stemming box in the search form to enable stemming for
all of the words in your search request. Stemming does not slow searches
noticeably and is almost always helpful in making sure you find what you
- If you want to add stemming selectively, add a ~ at the end of words
that you want stemmed in a search. Example: apply~
Synonym searching finds synonyms of a word in a search request. For example,
a search for fast would also find quick. You can enable synonym
searching for all words in a request or you can enable synonym searching
selectively by adding the & character after certain words in a request.
Example: fast& w/5 search.
The effect of a synonym search depends on the type of synonym expansion
requested on the search form. Keyword Search can expand synonyms using only
user-defined synonym sets, using synonyms from Keyword Search's built-in thesaurus, or
using synonyms and related words (such as antonyms, related categories, etc.)
from Keyword Search's built-in thesaurus.
Fuzzy searching will find a word even if it is misspelled. For example, a
fuzzy search for apple will find appple. Fuzzy searching can be
useful when you are searching text that may contain typographical errors, or for
text that has been scanned using optical character recognition (OCR).
There are two ways to add fuzziness to searches:
- Enable fuzziness for all of the words in your search request. You can
adjust the level of fuzziness from 1 to 10.
- You can also add fuzziness selectively using the % character. The number
of % characters you add determines the number of differences Keyword Search will
ignore when searching for a word. The position of the % characters
determines how many letters at the start of the word have to match exactly.
ba%nana Word must begin with ba and have at most one difference between it and banana.
b%%anana Word must begin with b and have at most two differences between it and banana.
Phonic searching looks for a word that sounds like the word you are searching
for and begins with the same letter. For example, a phonic search for Smith
will also find Smithe and Smythe.
To ask Keyword Search to search for a word phonically, put a # in front of the word
in your search request. Examples: #smith, #johnson
You can also check the Phonic searching box in the search form to
enable phonic searching for all words in your search request. Phonic searching
is somewhat slower than other types of searching and tends to make searches
over-inclusive, so it is usually better to use the # symbol to do phonic
Variable Term Weighting
When Keyword Search sorts search results after a search, by default all words in a
request count equally in counting hits. However, you can change this by
specifying the relative weights for each term in your search request, like this:
apple:5 and pear:1
This request would retrieve the same documents as apple and pear but,
Keyword Search would weight apple five times as heavily as pear when sorting the
In a natural language search, Keyword Search automatically weights terms based on
an analysis of their distribution in your documents. If you provide specific
term weights in a natural language search, these weights will override the
weights Keyword Search would otherwise assign.