NEJM Journal Watch search is a state-of-the-art tool designed for sophisticated collections of medical information. When a user searches for information, the search engine looks for not just literal matches, but matches with relevant metadata (i.e., medical terms and their synonyms). The result is that content, although written by multiple authors and derived from many resources, is completely integrated and connected by these topic-based medical language tags.
Principles of the JWatch.org Search: Beyond Full-Text and Word Matching
- JWatch.org search results do not rely on the exact matching of words as they appear in the full-text content, but on the true meaning of the concepts detailed in the content. Users can find all relevant discussions of the topic they are looking for, even if they are named differently in a multitude of resources, or if a common abbreviation is used in the search query.
- JWatch.org search results are listed in order of "relevance"; articles whose metadata matches the search terms and whose text contains the greatest number of the search terms and their synonyms will be listed first (but subject to a relevance test -- see below). Search results can also be sorted by date by the user.
- JWatch.org search is designed to strongly limit the number of irrelevant results. For example, if the user searched for information about "migraines," a search result linking to a discussion beginning with the sentence "Cluster headaches, unlike migraines, may occur in patients with..." would appear nearer the end of the results list since that content is actually not about migraines.
- JWatch.org search results display the metadata (tags) for each article as "related searches." Thus, for example, users may easily move from a results list of articles about "diabetes" to a results list for articles about the related topic of "glucose intolerance."
Be sure to check the spelling of search terms if the result of a search is invalid. You may be able to improve your search results by using the "wildcard" (*) if you know how to spell the beginning of a word but not the whole word.
It is possible to bypass the semantic programming described above. For cases when your requested information is very specific, the following search utilities are available:
- Phrases and Titles — Enclosing a phrase or title in quotation marks (" ") will narrow the results to just the desired phrase/title. For example, entering the phrase "breast cancer" in quotation marks will focus search results on that whole phrase, not the two individual terms. Entering the quoted term "HIV testing and counseling" will return a specific article or set of articles that have that exact phrase in the text — or that are directly relevant.
- Searches with "and" between search terms will generate results for articles that contain both terms.
Example: diabetes AND hypertension
If you enter this query, you will retrieve only articles that mention both diabetes and hypertension.
- Searches with "or" between search terms will generate results for articles that contain either term.
Example: diabetes OR hypertension
This query ensures that you get results for articles that include either diabetes or hypertension.
- The wildcard character (*) can be used to search the beginning fragments of words, forcing a match with any word containing a given root. For example, a search for child* will return articles containing child, childcare, and children; likewise, a search for phospha* will return articles containing phosphate, phosphatase, and hypophosphatemia. Wildcards can also be used to truncate words before non-English characters such as an umlaut (ü) or an accent (é). Since these characters cannot be searched, a word such as the author name Glück should be searched as Gl*.
Do not use a wildcard with a phrase; only use it for one word entries.