Hi,
I think there is an ongoing effort to add a full-text indexing feature to XIndice (you can search the archive of the dev ML for further details).
However, when I needed to implement a full-text search engine for an XIndice application, I used Apache Lucene,
http://jakarta.apache.org/lucene/docs/index.html
I simply used XIndice to store the full document bodies, and Lucene to build, store and query the index for the texts, along with a special "field" (the term for Lucene entires' slots) for each document, specifying the document path into XIndice (collection and id). So now I can run a query using Lucene, obtainig a list of XIndice document ids.
Francesco
At 20.13 16/06/2002, you wrote:
Hi,
Is anyone working on adding Full Text support (indexing and queries) to xindice ?
I have a 60,000+ record database that need revamped. It isn't currently XML but I think it would be ideally suited to becoming an XML database. Performing updates using XUpdate would be fantastic. Being able to execute XPath/XQuery queries would be nice but for most users the need for a simple, high performance keyword search is essential.
The data is currently imported into a BRS/Search database but our licence for that will expire in a couple of years and the University is unlikely to continue to pay for it.
A possible temporary would be to have all the data stored and updated in an xindice database which is then exported to a full text database for keyword searches. Are there any Free high-performance full-text databases out there ? I find it quite hard to assess their performance without actually going to the effort of trying them out. I tried ISearch which initially seemed like a good solution but it performed very badly when doing a join on a very common keyword and an uncommon one.
Cheers,
nick.
- Francesco Bellomi "Use truth to show illusion, and illusion to show truth."
