Searching Indexes
and Databases
Searching for information on one particular topic is relatively
easy: just look up the term you are interested in. But if you want to
find sources that discuss a particular set of different topics, it
can be time consuming to look up and make lists of all the sources on
each individual topic and then pick out those sources which are on
all the lists. Also, not being familiar with the specific vocabulary
or standard subject terms used for the topic can present difficulties
when you do not know what terms to look up (e.g. does the index use
"elderly," "senior citizens," or "aged"?). But there are a few tricks
to get around these problems when searching for information on a
computer database.
Here are some common searching techniques which offer powerful
tools for quickly and accurately broadening, narrowing, or refining
your search on a computer database. Most online indexes and databases
have at least some of these search options, but there are often
differences in how to use them, so read the help screens for the
database you are using.
Precision and
Recall
But first, a little searching theory, to see how these techniques can be
useful. When you search an online database, you want to find as much
relevant information in the database as you can while avoiding
getting irrelevant items along with your results.
The
"precision" of a search result refers to how much of the information
your search retrieved is actually relevant to your needs. Precision is easy to
determine: for example, you may have 40 records retrieved by your search, 30
of them are useful, the other 10 irrelevant. If your search retrieves
a lot of irrelevant results, you can use some of the techniques below to make a
more precise search which will retrieve less irrelevant material.
"Recall" refers to how much of the relevant information in the
database you were able to find with your search. This is much more difficult to
determine: you know you found 30 useful items, but you don't know how much useful
information you didn't find. The database may contain another 100 useful items
that you did not find, but you wouldn't have any way of knowing that for sure.
By using some of the techniques below to increase recall, you can be more
confident that you have found at least most of the useful information in the
database.
You can increase precision by being very specific and narrow
in your search, which will give you few if any irrelevant results, but
that risks missing a lot of useful information. You can increase recall by
broadening your search and thus finding more useful information, but that will
tend to increase the "noise" (irrelevant results you have to sift out)
in your results. So, increasing one tends to decrease the other. But you can use
the techniques below to find ways to maximize both precision and recall as much
as you can. Also, consider the purpose of your search. If you are using an
Internet search engine to find a few good web sites on a topic, focus on
increasing precision: it doesn't matter if you don't find everything available,
and you don't want to have to wade through thousands of irrelevant sites. But if
you are using an academic database to do a literature review for an extensive
research project, you want to make sure you find as many of the relevant
articles as you can, so you would focus on increasing recall even if you have to
sort out a lot of irrelevant articles in your results.
Keyword
vs. Subject Searching
Doing a "Subject" search searches only assigned subject
headings. A "Keyword" search will search titles, subjects, abstracts,
etc., and so is a broader, though less focused, search (i.e., it will likely retrieve more
results, but will also likely include some less relevant results and miss some
useful information).
|
 |
|
Keyword: find any record with "shakespeare" in
the title or contents, or as an
author or subject. |
Subject: find only those records
with "shakespeare" as a
subject. |
Concepts can often be described in different ways and with
different terms. Many indexes and databases use a standardized list
of subject terms to describe topics (called a controlled vocabulary).
This is very helpful if you know the standardized subject heading,
because you can do one search to find everything in the database on a
particular topic rather than having to do several searches. For
example, if you know a database uses the term "aged" as a
standardized subject heading, you can do one search for the subject
"aged" and know that you have also retrieved articles on "senior
citizens" and "elderly." Some databases will offer a thesaurus to
direct you to their chosen subject headings.
| If you do not know the appropriate standardized subject
terms for your topic, do a keyword search first. With a keyword search, you
use your own words or phrases to describe your topic; the computer will find
all the citations in the database that have those words anywhere in their
titles, subject headings, abstracts, or full text. Keyword searches often
turn up too much information, and sometimes much of it will be unrelated to
your topic. When you find a few appropriate citations, use their subject
headings to do a more focused subject search to locate all items with the
same subject headings.
 For
example, if you search in an online library catalog that uses the Library of
Congress subject headings and look for books on "cultural evolution," a
keyword search would retrieve this record for the book Cultural
Evolution: Contemporary Viewpoints. But note that "cultural evolution"
is not listed in the subject headings. A subject search would not have found
this book. But a subject search for "social evolution" will find all books
with titles about "cultural evolution," "cultural change," "social
development," and other synonyms. |
Note, however, that some databases (typically, those which do not use a list
of standardized subject headings) use the term "subject" search when
they really mean a "keyword" search. Knowing the type of search you
are actually doing can help you understand why you got your search results and
how to use those results (e.g. if there is no real subject search, you may have
to use other search techniques to make sure you have found all the relevant
information available in the database).
Broader
/ Narrower / Related Terms
If you do not find enough information on your topic, think
of synonyms or other ways of stating your topic and do another
search. Or, use more general terms to broaden your search.
If you find too much information on your topic, use more
specific terms to narrow your search, or use synonyms or related
terms to refocus your search.
|
Databases that use subject headings often include a
thesaurus of broader, narrower, and related terms for their
subject headings.
For example, looking up the subject heading "visual
memory" in PsycInfo's thesaurus will give you this list of
broader, narrower, and related subject headings which you
can use to expand, narrow, or refocus your search. Click on
any of the terms to find definitions and broader, narrower,
and related terms for them also, or select one or more to
search in the database. |
|
|
|
|
Boolean
Searching
Use AND, OR, and NOT operators to combine two
or more search terms. This allows you to define a complex set of
search criteria in a single search. By using the Boolean operators
AND, OR, and NOT, you can refine and specify your search to find just
what you want.
AND - Finds records in the database with all of your search
terms; i.e., AND narrows a search and makes it more specific by
allowing you to find sources which are about both of two different
topics.
Example:
|
|
Find only those records in the database which
include both terms.
|
OR - Finds records in the database with at least one of
your search terms; i.e., OR broadens a search by allowing you to find
sources which include one or the other (or both) of two different
terms.
Example:
|
|
Find all records in the database which are
about either Athens or the Olympics, including those
sources which are about both Athens and the Olympics.
|
NOT - Excludes records which have a specified search term
from your results; i.e. NOT narrows a search by allowing you to
eliminate a subset of sources from a larger set.
Example:
|
|
Find sources on anything about Athens except
for things about the Athens Olympics.
|
Complex Boolean searches - You can make very complex
searches by using more than one Boolean operator in a search:
Example:
 |
Starting from the innermost parentheses, find anything about
Athens or Sydney; then limit that set to anything also about the
Olympics; then discard anything about boxing. |
| Note that the placement of parentheses is very important to
make sure you get what you are looking for: complex Boolean
searches start from the innermost parentheses and work their way
out. |
Truncation
An easy way to broaden your search by searching for variant
endings of a word. You truncate the root of the word and search for
all the variant endings of that root.
For example, to search for information on computers and
computerization, type
"comput*." You will retrieve
everything in the database on computers, computing,
computerize, computerization, etc. Some databases also have
"wildcard" symbols, to search for variant spellings within a
word. For example, "lab*r" would find 'labor' and 'labour'.
|
 |
comput*
finds
computer
computers
computing
computerize
...
|
Note: Different databases use different keys for
truncation and wildcards (usually * or $ or ?) so check the help
guide for each database you use.
|
|
|
|
|
Choose the most efficient place to truncate a term, i.e., do not
truncate it such that your search will either retrieve a lot of
information you are not interested in or leave out relevant
information. For example, searching for
"com*" would retrieve information on
computers, computing, etc., but it would also retrieve information on
such topics as comedy, comets, compasses, and commodes. On the other
hand, "computer*" would retrieve
computers and computerize, but not computing.
Phrase
When you enter two or more terms in a search box, many search
tools will default to a Boolean "AND" search and retrieve those
documents which have all the terms somewhere in the document. Others
default to a Boolean "OR" search, finding those documents which have
at least one of the terms. Phrase searching searches for two or more
words as one search term rather than as two or more individual words.
Most databases allow phrase searching by putting quotation marks
around the phrase.
For example, if you are searching for the United Nations
Population Fund, it would be more efficient to search it as a phrase
rather than as individual words, since the terms "population" and
"fund" are likely to show up somewhere in other documents about the
United Nations. To narrow your search to find only those documents
which mention the United Nations Population Fund specifically, use a
phrase search:
Phrase searching is also useful when your phrase contains "stop
words," common words such as 'a', 'an', 'the', 'of', etc, which are
usually ignored by search tools. By using a phrase search, you can
search for the entire phrase, stop words included. Thus, you can
search for a phrase such as "state of the art" to find all uses of
that phrase in the database.
Proximity
Specify that two or more terms must appear close to one another,
e.g. "adjacent" to each other (in any order), "within" the same
subject heading, sentence, or paragraph, or "near" (e.g. within 5
words of) each other. This is especially useful for searching full
text databases.
|
|
This search will limit your
results to articles with 'diabetes' and 'children' within the same
sentence, thus
avoiding articles that talk about diabetes in general in one
part and mention children in another. |
|
|
|
Specific commands for proximity searching vary between databases,
so consult each database's help screens to use them properly.
Item
Type
Some databases allow you to narrow your search by specifying just
one type of source to find, e.g. books or journal articles or sound
recordings. Another option in some databases is to limit your search to
find only peer reviewed articles, i.e. articles in academic journals which are
reviewed by other scholars and researchers in the field before they are
published to check for accuracy and importance of contents.
|
For example, this index allows you to open a pull-down menu of a long
list of specific publication types. |
|
|
|
|
