2011/5/24 florent andré <[email protected]>: > Thanks, > > I see this https://issues.apache.org/jira/browse/STANBOL-209, but seems not > related to my problem. > > Nothing appear in the logs, but I figured out that on the very first > process, this work. > On the second this fall down. > > I'm a bad guy (sometimes :) ), and in my UIMA engine I just put all the uima > analyse result in the content item metadata graph... and for all sofa finded > all the text is copied in the node... > > I little bit too much data I think. Will clean this and keep in touch. > > By the way, do you have some input about the life cycle of a Content Item ? > When this object is destroy ? At the end of the process pipeline ?
Whenever you want :) For the /engines HTTP endpoint it's short-lived. For the /contenthub HTTP endpoint it's meant to be stored in a triple store + in memory hashmap for the binary attachement part. In the longer term we plan to setup a solr instance to index the analyzed content items for fulltext indexing with faceted browsing. -- Olivier http://twitter.com/ogrisel - http://github.com/ogrisel
