IBM open sources enterprise search
Takes unstructured approach
IBM is open sourcing a jointly developed search architecture with a view to creating a common industry approach to querying unstructured enterprise data.
The company is today expected to announce plans to open source the Unstructured Information Management Architecture (UIMA) used in the company's WebSphere Information Integrator OmniFind Edition, WebSphere Portal Server and Lotus Workplace. IBM is also open sourcing the UIMA toolkit.
UIMA provides a framework for software tools and services capable of conducting context-based searches across millions of unstructured records, databases, content repositories and email systems.
UIMA was developed by IBM Research with "significant" input from the Defense Advanced Research Projects Agency (DARPA), along with other contributors.
Nelson Mattos, IBM distinguished engineer and vice president of information integration, said UIMA could return hundreds of relevant documents from a search query compared to a key word search-based approach that would return millions of documents.
According to Mattos, IBM hopes to create an industry standard for text analytics through the release of the code. "The key goal is to create a forum for other research institutions to contribute to and develop the framework without having to depend purely on IBM to support it," he said.
IBM also hopes to attract buy-in from the commercial sector. As such, IBM is today also expected to announce 15 companies will use UIMA as the framework for planned search and text analysis tools.
Open sourcing UIMA is the first-step in a process that could see IBM adopt existing industry standards for use with the architecture. IBM said it would investigate use of the Object Management Group's (OMG's) Unified Modeling Language (UML), eCore, and XML Metadata Interchange (XMI) with the UIMA's Common Analysis Structure (CAS) specification later this year. CAS handles data exchange between UIMA's various components. ®
Sponsored: Hyper-scale data management