Knowledge-Aided Interface for Big Data Streams
Abstract
This SBIR Phase I project explored a knowledge-aided interface for Big Data streams, specifically considered in the context of multimedia signal processing for real-time alerting capabilities. The effort successfully demonstrated: a flexible methodology for audio and video signal collection on low-powered computing devices; a comparison of several stream data processing frameworks, resulting in the selection of the Amazon Kinesis service; state-of-the-art speech recognition and keyword search, as validated in the NIST OpenKWS evaluation and competitive with performers in the IARPA Babel program; design and initial prototypes for semantic knowledge integrated with context-sensitive visualization in a novel user interface. The technology developed is capable of capturing live broadcast news sources such as streaming Internet video. This is implemented using desktop operating systems that can run on very low-powered computing hardware (typically less than 10W power consumption).
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 29, 2014
- Accession Number
- ADA613209
Entities
People
- Arlo Faria