Text Mining and Analysis
The NCHC developed technologies that can process multiple data types for big data analysis in the cloud, in which “text” is a type of non-structural data. Text mining is a use of text analysis technologies to explore interesting information and useful knowledge in text data, and provides support for decision making. Furthermore, technologies and models required for applications are developed through statistical approaches, natural language, and machine/deep learning. The NCHC uses its high-performance computing and high capacity index storage structure based on these methods to develop vocabulary vector generation, text annotation, text classification, association analysis, opinion analysis, and event detection technologies. The technologies have been applied to analyses of disaster information, abnormal record, social network, and issue detection and analysis, along with food safety issue monitoring
- Scientific Breakthrough
- Overcomes traditional text-mining technology's constraints on semantic association and provides higher quality semantic association, expanding text mining technologies and applications to include text selection, classification, and event detection.
- Uses machine learning to explore associations in social networks and for issue detection and changes.
- Analyzes equipment logs and effectively detects and predicts abnormal events.
- Applies deep learning technologies to image object analyses and visual captioning, and analyzes dynamic behavioral models in images.
- Industry Applications
Provides industries with services to effectively process and store non-structural text, which is used for subsequent application and analysis. Text mining technologies can help industries effectively manage their document data, system logs, equipment monitoring records, and repair records and enables abnormal event detection and prediction, as well as maintenance cycle prediction and analysis through data mining.
Contact Person：Mr. Tseng