The following techniques can be used to increase your recall. Not all of the technqiues below can be used in every internet or library database; this listing let's you know what might be available. Of course, it's best to check the database's help file to determine if the technique is supported.
In most traditional library databases, you need to specify how the words you input relate to one another. All of these traditional databases understand the AND and OR operations.
Basically, the advice for the Boolean operator "NOT" (in some databases "ANDNOT") is ... do NOT use it! The NOT operator works well for high precision searching but in high recall searching you will very likely eliminate items you want from the results.
If you are tempted to use a NOT operator to eliminate unwanted material from a large result set - reverse your thinking. Think of terms you DO want to see in the results set, then AND those terms into your search statement.
Words take different forms or spelling depending on their use - nouns can be singular or plural, verbs change with the tense, and there could be adjective, adverb, or gerund forms of the word as well. (And don't forget variations in spellings between British and American English.) For high recall results, the searcher must account for these variations. Fortunately, most search interfaces have ways to help with this issue; unfortunately, how this is done is not consistent.
You'll need to consult the search interfaces's/database's help file; look for a section labelled truncation, wildcards or autostemming. These are the different ways search interfaces might handle word variation:
Beware the collective noun, for it does not retrieve all that you think!
If you use a collective noun in your search statement, you will also have to do a separate search for every item covered by that collective noun.
It's very easy to forget that bibliographic databases are actually searching the sequence of letters and spaces you enter in the search box rather than whole words and the meanings behind those words. As humans, we tend to group like things together and call the group by a name (ex.,fruit). We know that fruit means the group of food items that includes apples, oranges, pears, bananas, watermelons, kiwi, etc. But to the database, "fruit" is nothing more than the letter "f" followed by the letter "r" followed by the letter "u" and so on. What is retrieved are only those records that have f-r-u-i-t in them - you may see articles about apples and oranges in the results list but that's only because the word "fruit" is there, too. What you won't see are articles about a specific fruit (ex. apples) in which the word "fruit' does not appear.
Some databases have ways to help with this:
When you are looking for synonyms and other terminology for your concepts, don't forget to think of the opposite. For example, you may describe the concept as "safety" but others could think of it as "preventing accidents or fatalities".
Having trouble with very large answer sets? You may add another concept to your search, but only one; either use the C (comparison) concept or the O (outcome) concept but not both at once. Putting all four PICO concepts in your search statement usually results in very little, or no, results.
Not every database searches the full text (all the content) of documents; sometimes the database only has the citation (title, author, source) and an abstract (summary) to search through. You may need to adjust your search stategy depending on the searchable content of the database.
For databases that do not search full text, if your original strategy seems to retrieve too few results, consider using a broader version of your concept(s).
For example, consider this book ...
Author: Reier, Sharon.
Title: The Bridges of New York.
Source: Quadrant Press
One of the 75 bridges discussed in this book is the Williamsburg Bridge. If you searched Google Books (which indexes the full text of books) for Williamsburg Bridge New York, you would find this book in the results list. This book is also in the WorldCat database, however, as only the citation information is contained in WorldCat, the search for Willamsburg Bridge New York would not retrieve the book.
When searching in a citation only (or citation with abstract) database, keep in mind the possibility that any or all of your concepts may be just a small part of a larger (i.e., broader) concept. If you want highest recall results you may need to search the broader concept as well. For the Williamsburg Bridge example, that means you would also search for bridges AND New York as well as Williamsburg Bridge.