compared with
Current by Reneta Popova
on Dec 08, 2014 15:27.

Key
This line was removed.
This word was removed. This word was added.
This line was added.

Changes (1)

View Page History
h1. Architecture features

As the Semantic Extraction Service is wired to GraphDB through its Plug-in API, its dictionaries are dynamically updated in a transactional fashion and are always in synchronization with the data in the semantic repository. It also provides high availability ensured by GraphDB's replication, load balancing and fail-over capabilities. Horizontal scaling also works in the same way as in GraphDB. Moreover, the pipelines running on a single cluster node could be pooled in the way so that multiple pipeline instances share the same dictionary and achieve more semantic annotation request throughput while optimizing memory usage. The pools are also capable of dynamically expanding with respect to the current load. Communication with the service is established over HTTP and the SPARQL protocol, which makes it easily integrable with various platforms written in different programming languages.

h1. Interfaces, Standards, Requirements