David Milne - 2013-12-18

Hi Guys,

This is due to label caching. In the web services we don't use the database directly. instead we cache the vocabulary of labels to memory (if we didn't, the annotate service would be painfully slow).

We don't cache all labels, but only those that have a reasonable probability of being a link. Unfortunately 'Learning' is found in many, many sentences where it is not used as a link, so it has a very small link probability and doesn't get cached.

I have an idea for how to fix this and have logged an issue. You can keep tabs on it here: https://github.com/dnmilne/wikipediaminer/issues/13