Rob Koopman wrote:

> Ray wrote:
>
> >which may include:
> >dc.titleAllTheseWords
> >dc.titleAnyOfTheseWords
> >dc.titleAdjacentWords
> >dc.titleRelevantWords
> >dc.titleString
> >
> >I don't think dc.titleWord will be one of these.
>
> Maybe I am out of sync but I do not really understand the purpose of these.
> Does this mean we write:
> query = (dc.titleAllTheseWords="a b c")?

Yes.

> My logic tells me that this is mixing index type with default operator,
> resulting in an explosion of pseudo index types all translating to word.

For better or worse the ZIG has adopted all OfTheseWords and AnyOfTheseWord as
Utility set attributes (currently format/structure but I think misplaced and
should be expansion/interpretation) as a convenience to avoid constructing
booleans. This debate has lingered for over 10 years but it seems that the
convenience faction has won.

I do think we all agree we need  adjacentWords, and I do think a case can be
made for relevantWords   -- for adjacent words, true, you could construct a
boolean query with proximity but the argument is that it's unlikely to be widely
supported;  for relevantWords, you can't represent that as a boolean query.  So
it follows that we need some sort of word list, and it follows from that that we
need some way to interpret the word list (adjacency or relevance).  So, it's not
too much of a leap to suggest an additional two types for All and Any.

--Ray