Service Learner
This module extracts terms and relations ( i.e. hypernym, mereology and domain relations) from the documents to create a lexicalized ontology for that domain.
It uses the Natural Language Processing (NLP) service to perform extra annotation tasks, not previously executed by the harvester module, e.g dependency tagging, and the Knowledge-Base service, which was previously trained using a large domain‐independent corpus of text and one or more knowledge bases.
These are the routing-keys from the event-bus that this module is listening for or publishing to:
domain.{closed}
relation.{new|opened|closed|analyzed}word.{new|opened|closed|analyzed}
When a new notification has been received in the domain.closed channel, this module will read all the documents in that domain to extract terms and relations, and to build a taxonomy from them. For each entity recognized, it will publish a new event-message to the word.closed or to the word.new channel, depending on they are the same word or a composition of words.
After that, it will gather all the relevant information about the text using the NLP service and the Knowledge-Base service to conclude what is the type of relation that exists, or not, between two terms. For each new relation discovered, a new event-message to relation.new or relation.closed channel will be published.
