ECCParaCorp

ECCParaCorp provides a parallel English–Chinese corpus of cancer-related texts aligning 6,685 text pairs across six cancer types and three thematic areas to support cross-lingual biomedical natural language processing and knowledge extraction.

Key Features:

• Corpus composition: ECCParaCorp comprises 6,685 aligned text pairs in XML, Excel, and CSV formats, including 5,190 sentence pairs, 1,083 phrase pairs, and 412 word pairs.
• Cancer coverage: The corpus covers breast cancer, liver cancer, lung cancer, esophageal cancer, colorectal cancer, and stomach cancer.
• Thematic coverage: Texts are organized across the thematic areas of prevention, screening, and treatment.
• Source: Content is drawn from authoritative cancer information in PDQ (Physician Data Query).
• Cross-lingual alignment: Aligned English–Chinese text pairs provide structured cross-lingual information for mapping terms and sentences between languages.

Scientific Applications:

• Machine translation: The parallel corpus can be used to train and evaluate machine translation models for medical scenarios.
• Educational system development: The aligned texts support cancer-related system development for educational purposes.
• Knowledge extraction: The corpus facilitates disease-oriented knowledge extraction from bilingual texts.
• AI model training from clinical text: The data can be used to enhance AI model training from non-structural clinical narratives and electronic health records.
• Medical education and public health literacy: The corpus can support efforts to improve medical education and public health literacy through bilingual resources.

Methodology:

The corpus was created via a seven-step workflow: data retrieval, parsing, processing, implementation, assessment verification, release, and application.