PADI-web

PADI-web monitors online news sources to detect and extract animal disease information for biosurveillance of emerging infectious diseases in animals.

Key Features:

• Fully automated pipeline: A continuous automated pipeline processes incoming news articles without manual intervention.
• Machine-learning methods: Advanced machine-learning algorithms are employed for data processing and classification.
• Fine-grained classification system: A classification module assigns detailed categories to extracted information to improve specificity.
• Automatic term and named entity extraction: Text-mining techniques extract epidemiological terms and named entities from news content.
• Semantic resources for indexing keywords: Semantic resources are used to index and organize keywords for retrieval.
• Configurable notification system: An automated alerting mechanism generates configurable notifications based on detection criteria.
• Multilingual processing: The system processes news in multiple languages to broaden geographic and linguistic coverage.
• Corpus of news articles: The platform's database contains over 380,000 news articles for analysis.
• Integration with ESA Platform: PADI-web is integrated within the French Platform for Animal Health Surveillance (ESA Platform).

Scientific Applications:

• Animal disease surveillance: Detection and monitoring of emerging infectious diseases in animal populations using online news sources.
• Epidemiological information extraction: Extraction of disease-related terms and named entities to support epidemiological analysis.
• Keyword indexing and retrieval: Semantic indexing of keywords to enable organized retrieval of relevant disease reports.
• Multilingual biosurveillance: Surveillance across multiple languages to support regional and international monitoring.
• Integration into surveillance networks: Provision of processed signals for incorporation into national or platform-level surveillance systems such as ESA.

Methodology:

Automated processing of online news using a fully automated pipeline that applies machine-learning methods, text-mining techniques, fine-grained classification, named entity extraction, semantic resources for keyword indexing, and multilingual processing on a corpus of over 380,000 news articles.