This section will explain the features of the search interface. These include single word search, phrase search, Boolean operators search capability, proximity and stemming search options.
A search for a single word is the most straightforward way to perform a search. The application will look through all documents in the database, and retrieves documentss that contain the search term.
Searching for a phrase is straightforward as well. A search for a multi-word phrase, such as breast cancer risk, will return clinical trials that contain all words in the phrase. The default behavior searches as if AND was added between each word of the query. Enclosing a queried phrase in quotation marks requires all words in the phrase to be found adjacently, in the exact order, in a document to be considered relevant. For this reason, the search “breast cancer risk” is not equivalent to searching for breast cancer risk.
Users can use Boolean operators to refine their queries as well. Boolean logic can be added to any free text query by adding AND, OR, or NOT to the query. As stated above, the application defaults to adding AND between all words in the query. Using NOT will eliminate all documents that contain the negated term even if they contain the first word of the query. The addition of OR between two words or phrases in a query relaxes the search restrictions of AND. If OR is used, the application not only finds all occurrences of the full phrase, but also finds each word individually. Higher relevance is then given to documents that contain the full phrase.
It is important to note that AND differs from using quotation marks. AND does not require the terms in the query to be adjacent in a document; quotation marks do require that the terms in the query are adjacent to each other. For example, a search for “breast cancer” is not the same as a search for breast AND cancer. The two methods for refining a query may be used together though, e.g. “breast cancer” AND taxotere is a valid search.
Proximity searches can be performed by using either the tilde (~) or percent (%) operator. A proximity search is used to find terms within a specified range of each other. The syntax for this type of search is “term 1 term 2” ~ “range”. For example, a search for “lung cancer”~5 will look for lung and cancer within 5 words of each other. This could be useful if lung was found in a list of cancer types (e.g. the above query would find a match if the document contained the following list: lung, breast, and colon cancer).
Users can also use an asterisk to search for all terms that begin with the same part of a word, for example, asthma* will find documents containing asthma, asthmatic, asthmagen. (Please note that the current implementation does not support wildcard inside phrase searches.)
Learn More