Hi Pablo (cc to the mailing list),

sorry for the long delay; here's finally my detailed answer. I spent some time in revising the "AnyAllOptions" (any word, all words, phrase, ...), and I noticed that not all combinations delivered the expected scoring results. The issue should now be solved; feel free to tell me if you need a jar build of the latest build; or check out the latest version from the repository.

Hello Christian, thanks por you prompt response. Enclosed I send you the Comuni file for Italy, that we use to perform the math of others 32000, with the final purpose of validate its. […]

Thanks for reporting on your usage scenario. One way to speed up your queries might be to create a full-text index (note, however, that the AnyAllOptions are not supported by the index). The improved performance gain after the first call, which you observed, might be due to the caching of results. To get better performance, I'd recommend to switch to the client/server architecture of BaseX. Apart from that, it's difficult to give you general guidelines on how to handle your amounts of data. Some questions might have to be answered first, such as…

– do you plan to store the data and perform repeated queries, or will you just process it once/on the fly? – what's the total size of your XML documents? how many documents/XML snippets will you look at?

Hope this helps, Christian