searchrefiner Help

home

Contents

  1. Writing Queries
  2. Loading PMIDs
  3. Using the Structured Editor
  4. Using the QueryVis Tree View

Writing Queries

Currently queries can be submitted to searchrefiner in two query languages. The first query language is Ovid MEDLINE. these queries look like this:

1. exp ORTHODONTICS/
2. orthodontic$.mp.
3. or/1-2
4. (retention or retain$).mp.
5. (stabilise$ or stabilize$).mp.
6. (fraenectom$ or frenectom$).mp.
7. (fiberotom$ or fibreotom$).mp.
8. interproximal stripping.mp.
9. pericision.mp.
10. reproximat$.mp.
11. ((gingiv$ or periodont$).mp. adj4 surg$).mp.
12. (retain or retention).mp.
13. 11 and 12
14. or/4-10
15. 13 or 14
16. 3 and 15

In the Ovid MEDLINE query language, keywords of the query can be expressed individually on lines (e.g. line 1 or 2), or grouped together (e.g. line 4). These keyword clauses can be combined with infix logical operators if grouped on a single line or by stating which lines to combine (e.g. line 15 combines the clauses specified on line 13 and 14 with the OR operator). Keywords are restricted to fields by the two letter identifier at the end (e.g. .mp., which restricts the keyword to the title, abstract, and MeSH headings). MeSH only keywords are specified with / and can be exploded with exp (e.g. line 1). Keywords can also be explicitly stemmed or expanded with the $ and * modifiers.

For more information about building MEDLINE queries, see http://ospguides.ovid.com/OSPguides/medline.htm

In the PubMed query language, queries look like this:

((ORTHODONTICS[Mesh] OR orthodontic~[All Fields]) AND (((surg~[All Fields] ADJ4 (gingiv~[All Fields] OR periodont~[All Fields])) AND (retain[All Fields] OR retention[All Fields])) OR (reproximat~[All Fields] OR "interproximal stripping"[All Fields] OR pericision[All Fields] OR (retention[All Fields] OR retain~[All Fields]) OR (stabilise~[All Fields] OR stabilize~[All Fields]) OR (fraenectom~[All Fields] OR frenectom~[All Fields]) OR (fiberotom~[All Fields] OR fibreotom~[All Fields]))))

In PubMed query syntax, the query is contained on a single line, fields are indicated by square brackets and keywords are grouped with infix logical operators.

For more information about building PubMed queries, see https://www.nlm.nih.gov/bsd/disted/pubmedtutorial/cover.html

Loading Known Relevant PMIDs

Loading known-relevant PMIDs into searchrefiner allows one to identify how effective a query is at retrieving a baseline set of citations. To load PMIDs into searchrefiner, visit the settings page (also accessible from the sidebar). Type or paste in the PMIDs of citations that the query that is being formulated should retrieve. These statistics will then be visible in the QueryVis tool.

Using the Structured Editor

This is a two-way binding of textual query to a structured editor. This allows users to paste in existing queries (e.g. from other interfaces or existing systematic reviews) and edit them conveniently in a structured way by dragging and dropping clauses of the Boolean query. The logical operators, fields, queries, and MeSH explosion can be edited in this structured editor and all updates are reflected in the textual editor in real time. The structured editor can be used to supplement search interfaces in other tools because of this two way binding. For example, a query can be formulated in PubMed, but to make refinements to it without introducing syntactic errors one may paste it into the textual editor, edit in the structured editor, and copy it back to PubMed via the textual editor. The structured editor removes the requirement for understanding the details of such a query. It allows, in fact, novice users to write and edit queries without learning the different query syntaxes for each database that is used to retrieve literature for systematic reviews (all queries syntaxes look the same in the structured editor and the structured editor can output queries in different syntaxes).

Using the QueryVis Tree View

By using the tree view, one may visualise the number of citations that are retrieved by each clause of the query, and how the number of retrieved citations is affected by logical operators (i.e. AND, OR, NOT, etc.). This visualisation can be augmented with relevance assessments in order to ensure a monitor the retrieval of number of known relevant citations, or for training purposes.