CEC-Corpus

Chinese Emergency Corpus (CEC) Overview

Dataset Construction

• Construction Institution: Semantic Intelligence Lab, Shanghai University
• Data Source: Internet news reports
• Event Categories: Earthquake, fire, traffic accident, terrorist attack, food poisoning (5 categories)
• Number of Texts: 332 articles

Data Processing

• Preprocessing Steps: Text preprocessing, text analysis, event annotation, consistency checking
• Annotation Format: XML
• Primary Data Structures: Event, Denoter, Time, Location, Participant, Object
• Attribute Definitions: Define relevant attributes for each tag

Research and Development Funding

• Funding Projects: National Natural Science Foundation projects “Key Issues in Event Reasoning based on Description Logic” (Grant No. 61305053) and “Event Ontology Model and Application Technology” (Grant No. 60975033)

Research Outcomes

• Research Papers: Multiple papers published in Journal of Chinese Information Processing, Pattern Recognition and Artificial Intelligence, etc.

• Doctoral Dissertations: Including studies on event‑oriented knowledge processing and event‑oriented text representation.

• Master’s Theses: Covering intentional event research, extraction and reasoning of temporal event elements, etc.

Corpus Characteristics

• Scale: Slightly smaller than ACE and TimeBank corpora
• Annotation Completeness: Provides the most comprehensive annotation of events and event elements.