On 20.06.2013 23:27, Niall Gallagher wrote:
Hi Ceki,
Just wondering about the SUB-STRING patterns, such as "*/if/then/*" and "*/if/else/*".
I'm currently working on adding a permuterm tree to the concurrent-trees[1] project. It will support fast retrieval from the tree for wildcard queries like: X, X*, *X, *X*, X*Y. These patterns would be fully accelerated.
The *X* variant fits the sub-string patterns used in Joran. While traversing the elements of an XML document, I would like to compute if *X* is the path defined by current location in the XML document. This seems to be different (or even unrelated) to the problem of checking whether *X* matches one of the keys inserted previously into an ADT such as a trie.
It looks like the *X* variant might fit your sub-string patterns.
It's also possible to have partial acceleration for additional patterns such as X*Y*Z, *X*Y*Z, *X*Y*Z* and so on with any number of terms, which involves using the first and last terms to find a candidate set in the tree and then doing on-the-fly filtering for the terms in the middle.
I am curious. Is "autocomplete" in presence of typos the use case for such patterns?
Would that help? I'm about 80% of the way to completion.
[1] https://code.google.com/p/concurrent-trees/
Niall
-- Ceki 65% of statistics are made up on the spot