Apache UIMA

Apache UIMA is an open source, multiplatfomr and free Java SDK (Software Development Kit) and framework that includes all kind of tools and annotators, facilitating the analysis of unstructured content, such as audio, video and text.

UIMA stands for Unstructured Information Management Architecture, and are software systems capable of analyzing large volumes of unstructured data, in order to reveal knowledge that is relevant to an the end-user.

Supports Java and C++ programming languages

Apache UIMA supports both Java and C++. While the first one supports only Java and non-Java components, the latter supports annotators written in C, C++, Perl, TCL and Python programming languages.

The software comprises of the Apache UIMA DUCC, Apache UIMA uimaFIT, Apache UIMA Ruta, Apache UIMA Asynchronous Scaleout (UIMA-AS), Apache UIMA Java SDK and Apache UIMA C++ SDK components, as well as the UIMA Addons package.

Under the hood, supported OSes and availability

A quick look under the hood, will show us that the software is written in the Java programming language and it’s compatible with any operating system where the Java Runtime Environment (JRE) is available, including all GNU/Linux distributions and various UNIX-like systems.

For your convenience, all the components of the Apache UIMA project are available for download as pre-built binaries that can be used without any special configuration of installation, as well as archives that contains their source code. It is distributed as part of the Apache Software Foundation.

What is new in this release:

• The UIMA framework has been enhanced to support large memory/ large multi-core machines, with various improvements to synchronization, lock contention, and locality of reference to optimize memory bandwidth in the presence of non-uniform memory access speeds. In addition, embedders of the framework can now choose to run initialization of UIMA components that share a common Resource Manager and/or common UIMA Context, on multiple threads, concurrently. CASes which are not being modified can be accessed (read-only) by multiple threads, simultaneously. Along the way, many subtle bugs were fixed.

What is new in version 2.5.0:

• Replaced ActiveMQ version 5.4.1 with version 5.6.0
• Implemented PEAR support
• Added client-side callbacks to notify an application where a CAS is being processed
• Refactored JMS Session expiration management to prevent broker OOMs
• Refactored UIMA-AS Async Aggregate to support processing multiple input CASes in parallel
• Refactored CAS Accounting code to prevent loosing CASes in complex/nested deployments
• Improved error handling and recovery
• Performance improvements

Requirements:

• Java 2 Standard Edition Runtime Environment