DiSCVR

Team:  
Dakkak, Abdul; Pearson, Carl; Li, Cheng
     
Description:  

The Document Introspection System for CognitiVe Research (DiSCVR) is an extensible open source project that analyzes user-provided research papers and delivers recommendations for related content and papers.  DiSCVR is a pipeline of modular services, clients, and databases that extract Information about the research paper and match it with similar papers. DiSCVR provides 

  • A foundation for research into common programming, computation, and communication patterns of scale-out and scale-up cognitive computing applications.
  • A platform for implementation, integration, and evaluation of cognitive computing and learning techniques.
  • A realistic cognitive computing application that can be used to evaluate the performance and energy-efficiency implications of novel computer systems.
  • Compiler and scheduling opportunity for performing optimizations on application pipelines as well as scheduling services to run on ideal hardware architectures.
  • Insights from DiSCVR and other end-to-end cognitive workflows will help us develop a generic Cognitive Application Builder (CAB) to improve cognitive application developers' productivity and deployment scalability.