Apache UIMA - what is it?
Unstructered Information Management Architecture
Architectural Framework to manage (eventually large) volumes of
Former IBM Alphaworks project donated to ASF
Currently an Incubator podling ( http://incubator.apache.org/uima )
Apache UIMA is an Oasis standard ( http://www.oasis-open.org )
Apache UIMA - how?
Many pluggable reusable components (described via XML)
Analysis Engines (primitive or aggregates)
Asynchronous scaleout (JMS, Apache ActiveMQ)
Apache UIMA - what is NOT?
It’s not a semantic search tool inherently
the “Lucas example”
the semantic search package for UIMA is not open source!
( http://www.alphaworks.ibm.com/tech/uima/download )
UIMA & Semantic Search
Metadata generation engine for CM systems
Jeopardy (see http://www.research.ibm.com/deepqa/
RE Market Analysis & UIMA
Macpi: a real estate market analysis tool developed at DIA
Webpipe (crawling and wrapping data)
Extract metadata with Apache UIMA to build our search
Apache UIMA & AlchemyAPI
AlchemyAPI from Orchestr8 services wrapped as UIMA AEs
Named-entity recognition, word disambiguation
“Barack Obama” is http://dbpedia.org/resource/Barack_Obama
Exploiting linked data
enriching free text with DBpedia, GeoNames, Freebase URIs
Plugging with other UIMA AEs
providing you with a reusable component to deal with Linked Data