Searching in Confluence

This second post in a series on using Confluence effectively discusses using Confluence search.

(In my first post in this series, I discussed the virtues of Tiny Links in Confluence.)

This post draws from an excellent Kara Stiles post to the sakai-dev list which walked through effective searching using Jira. This post makes parallel suggestions for effective searching in Confluence.

Tips for effective Confluence search

Boolean search

Confluence supports boolean search.

For instance, suppose I'm trying to find the JA-SIG wiki page on CASifying uPortal. I may not be aware that this page is in the nascent uPortal Manual wiki space. A naive search for CAS uPortal doesn't yield this page in its first page of results.

This is in part because sometimes people write in terms of how to "integrate uPortal and CAS" and other times in terms of how to "CASify uPortal". If I don't know ahead of time whether the page will use "CAS" or "CASify", maybe it will help to find pages that include either "CAS" or "CASify", and "uPortal". Searching on uPortal AND (CAS OR CASify) yields four top results that are all relevant to effectively CASifying a uPortal instance.

Wildcard search

When the French CASify something, they tend to use the term "CASifier". Searching on uPortal AND (CAS or CASify or CASifier) may work but a wildcard better expresses the thought of wanting "CAS" followed by any characters. uPortal AND CAS* It turns out this introduces some "noise" in the search results from matches on "case", though, so wildcards aren't always more effective than enumerated terms in an OR.

Proximity

Suppose I'm looking for Matt Young's excellent presentation on Duke's revitalized uPortal effort. In an ideal world, this presentation would have its own page (look forward to a future post on object-oriented domain modeling in Confluence). In the present world, it doesn't have its own page. Searching for "Duke" and "Duke uPortal" doesn't find it. What to do?

Well, I bet that the page it's on has "presentation" on it somewhere, and while I don't know exactly the phrase involving "uPortal" and "Duke", there's gotta be one. "Matt" and "Young" are probably in there somewhere too, though I don't know ahead of time whether it's "Matt Young" or "Young, Matt".

While normal quotes match literal instances of the quoted text, the nearness operator softens the match to include cases where the words occur sufficiently near one another. For instance, "Matt Young"~3 should find Matt whether he is listed last-name-first or first-name-first. presentation AND "Matt Young"~3 AND Duke nicely matches the Duke presentation as the very first search result.

Wrapping up

In this quick post I've just touched a couple of Confluence search tips, the Confluence parallels to those advanced for Jira in the mentioned list post. Just the same, I too sometimes have better luck using Google to search Confluence, a which is conveniently quite feasible for spaces that are wide open to the public. Confluence's human-readable search-engine-friendly URLs are one of its most compelling features.

Resources for learning more

Atlassian maintains general documentation on searching Confluence as well as a specific page on search syntax including more discussion of boolean searching and the proximity operator.