To search for concepts and use proximity operators in databases

Databases most used at KI, are both databases with a controlled vocabulary (such as Pubmed and CINAHL) and databases that use free-text search (such as Web of Science and Scopus). There are quite big differences between these two types of databases, in terms of how to best search.

Pubmed uses a controlled vocabulary called MeSH. A search for a term in Pubmed maps automatically to the correct MeSH term (if any). You don’t need to know if the term is a mesh term or not, Pubmed will do it for you.

Always check Search Details, to see if your terms were mapped to mesh terms, or not.

If you search for a phrase or a concept, it can be tempting to put your search terms in quotation marks, to make sure it will be searched as a concept, for ex. "egg allergy". But don’t do that in databases that use a controlled vocabulary. If you do so in Pubmed, the automatic mapping that maps egg allergy to egg hypersensitivity [MeSH Terms] doesn’t work.  

How do you search for a term of two (or more) words, when the database doesn’t have a controlled vocabulary?

In a free-text database, the search methods differ compared to a database that uses controlled vocabulary.  It works just fine to search for a concept as a phrase with quotes: "egg allergy." This is important to do, when one of the words is a common word like disease, attack, allergy etc. If you search for a concept  like heart attack, it makes a big difference, in both relevance and number of hits, if you use quotes or not:

heart attack gives 24 030 hits
(Ambulatory heart rate changes in patients with panic attacks)

”heart attack” gives 5 148 hits
(Preventing heart attack and death in patients with coronary disease)

However, it makes no significant difference in either the relevance or the number of hits, searching for a specific concept like "rheumatoid arthritis" with quotation marks, compared with no quotes:

rheumatoid arthritis gives 134,998 hits
”rheumatoid arthritis” gives 133,686 hits

If you search for two terms in Pubmed, Web of Science or Google, the databases automatically add the Boolean AND between the terms. Then the database will find both of the terms somewhere in the abstract or title:

There is a third option, if you do not require that the words should be precisely sequential ("egg allergy"), but it is not enough that both words are found somewhere in the abstract or title (egg AND  allergy). Then you can use the so-called proximity operators. The database finds the words you search for near each other, but they do not have to be exactly next to each other, as in a phrase. Proximity is expressed in different ways in different databases:

For ex NEAR in Web of Science and adj in MEDLINE (OVID).

You can also determine how many words you "allow" between your two terms.

egg NEAR/2 allergy respectively egg adj2 allergy.

This means that you will find these two terms with a maximum of two words in between. A search for "egg allergy" does not retrieve egg and milk allergy, which egg NEAR/2 allergy does..

To see the proximity operator that applies in the database, you need to look in the help texts. It’s important to note that the proximity operators do not work in Pubmed.

Susanne Gustafsson (tjl.)

Bibliotekarie på KIB som träffar många av våra studenter och doktorander, både i undervisning och handledning.