Re: Xpath conundrum for the peeps

You're thinking of this as an XPath problem when it's actually a requirements
problem. Define the rules you want to implement, and we can help you express
them in XPath (or advise on a more suitable language).

Your choice of rules will depend very much on the quality and variability of
the data, and on whether you want to bias towards matching or not-matching in
cases of doubt. Those are your decisions to make.

Michael Kay
Saxonica
mike@xxxxxxxxxxxx
+44 (0) 118 946 5893




On 22 Aug 2014, at 08:31, Ihe Onwuka ihe.onwuka@xxxxxxxxx
<xsl-list-service@xxxxxxxxxxxxxxxxxxxxxx> wrote:

> Ok I admit I haven't thought about this in anger but a problem
shared....etc...
>
> I am doing a matching algorithm to match movie data from different
repositories so that I know when the repositories are referencing the same
movie even though they may hold different metadata.
>
> It's not enough to match solely on title - one reason for that is movies
have subtitles and may go by the subtitle in a different venue.
>
> So let's say thanks to xsl:key I have in a variable $titles all the movies
that have that title and in a variable $actors I have all the movies that that
actor featured in.
>
> A (out of many) criteria I could have is that if the data from the
respective venues has it's title and an actor in common then they are the same
movie, thats a plain intersect between $titles and $movies but I want
something stronger than that.
>
> I want the data from the venues to match only if they have at least 2 actors
in common.
>
>
>
> XSL-List info and archive
> EasyUnsubscribe (by email)