Jump to content

Building a complex search query


AdamS

Recommended Posts

I have a rather complex search which I'm not getting to work quite right so thought I'd come here for assistance.

 

Essentially this is what I'm trying to do, any instances of:

(approv* or verific*) and design, which appear withing 100 words of several other keywords.

 

"(approv* OR verific*) AND design AND (keyword1 OR keyword2 OR keyword3)" ~100

 

The above didn't work so I'm not sure if I can build the query I need this way or if I need to explode my head and try regex.

 

Any help appreciated.

 

Edit: interestingly if I remove the ~100 then the search query as it's structured will work. There's obviously some relationship issues here when trying to tell Intella what that word limit actually applies to.

 

Edit again: sorry, I just found the sticky by Jon which breaks it down nicely for me and I have figured out where my issues are.

Link to comment
Share on other sites

Even after reading Jon's very good post I'm still not sure on something, maybe I missed it.

 

Breaking down the complex search I still can't figure out how to use the proximity function when dealing with a phrase?

 

eg I want to find "beat" within 50 words of "here it comes". Using Jon's keyword method I search for both individually, then tag the overlapping so I have only documents which have both, however from here I'm still facing the same issue with not being able to apply a proximity search on a phrase.

Link to comment
Share on other sites

Hi Adam,

 

Because the double quotes need to encase all of the search terms, you cannot have a search phrase within a proximity search. A search phrase would require double quotes around it and you can't have nested double quotes within a proximity search. That said, you can use phrases in keyword or keyword lists to narrow down the results before running a simplified proximity search.

 

You have done the first part correctly (e.g. I want to find "beat" within 50 words of "here it comes". Using Jon's keyword method I search for both individually, then tag the overlapping so I have only documents which have both search terms).

 

Because you cant us a phrase in a proximity search, you need to run your proximity search for 'beat' and one of the words in the phrase e.g. "beat comes"~50. This could return false positives as 'beat' could be within 50 words of another word 'comes' which is not related to the phrase 'here it comes'. Unfortunately this is a limitation but this process will also return hits for 'beat' within 50 words of 'here it comes'.

 

I hope this makes sense.

 

Jon

Link to comment
Share on other sites

Thanks Jon, yes it makes sense and that was the approach I was taking. I'm lucky in this instance that one of the words in the phrase I need is quite unique.

 

Possibility for the roadmap maybe to allow nested phrase proximity searches?

 

Or alternatively if there is anyone on here who is a regex guru may wish to chime in ;)

Link to comment
Share on other sites

×
×
  • Create New...