PART TWO:
Fuzzy search tricks: This deals with scenarios where one or more letters could vary - this is very handy when the information you're looking for might have poor image quality, or words/letter may be hard to correctly recognise or be inconsistently spelt.
Note however that fuzzy searches don’t invoke the highlighting feature on the site. (fuzzy search for a term where any one character can mismatch)
(same, but two characters can mismatch)
- * = multi-character wildcard
- ? = single-character wildcard
Eg 1: jones~1 (this will give you matches for “Jones” even when the OCR has mis-recognised 1 letter, for example “johes”)
Eg 2: Hartstone~2 (matches for when any two letters might not be identical)
Eg 3: auck?and (matches where the ? can be any one letter)
Eg 4: auckl* (matches where auckl can be followed by any number of letters)
A cool trick with this is that searching for fuzzy results for a term and using NOT to remove exact character matches for the canonical term gives you a list of results containing only OCR or spelling variations for the term. This can be handy when a name first appeared in a country and may have initially had numerous spelling variations.
Eg 5: puhirake~2 –puhirake
Eg 6: aluminium~2 OR aluminum~2
Weighting operators: ^x assign x more weighting to relevance score in the ranking of a term in your search results.
This is really handy when you need to customise how results for two or more words are sorted, but sort by date or some other sorting doesn't cover what you need.
Grouping: Two techniques here, (parentheses), or “term1 term2”~x: Note the slightly different use of the tilde here than in the “fuzzy search tricks” section above: if you invoke it after a group of words in speechmarks, it expresses an n-gram length for word strings up to that length containing those terms in any order
This gives results when “Biscuit” and “Barrel” occur in any 4-word string. For example, results for this would include things like the phrase “barrel full of biscuits” (a 4-word string) but would not include “biscuits should never be stored in a barrel”, an 8-word string.
For example: (chicken +egg) OR "trucks with flatbed trailers" - you can have one search term with a boolean AND operator only applied to it by enclosing them in parentheses, and another part that avoids the boolean AND by being outside the parentheses.
Alrighty, these are most of the building blocks for our various search operators - I'd encourage you to explore them methodically if you can, be careful not to try to create "silver bullet" search terms that go straight for the jugular. Instead, start with the simplest, broadest search you can (so for genealogy, search for a surname), and then gradually introduce tweaks to your search term one change at a time, so that you can see the impact of each change in the results you get, and respond accordingly.
Being methodical and doing it step-by-step will keep you alert to unintended "collateral damage" in your search results as a consequence of mis-using boolean terms. If you ever get totally stumped at what you're seeing, just let me know (use the email address listed on the site under "contact us") and I'll help!