Overview of the NLPCC 2024 Shared Task 2: Nominal Compound Chain Extraction
摘要
Nominal compound chain extraction (NCCE) represents an emerging task within the domain of natural language processing, exhibiting significant potential for application in various downstream tasks, including relation extraction and summarization, etc. This task involves analyzing a document to identify and categorize all nominal compounds into distinct clusters based on semantic relatedness. Consequently, NCCE presents greater challenges than traditional coreference resolution tasks due to its reliance on semantic associations. In response to these challenges, we initiated a shared task at NLPCC 2024, attracting 15 teams to register, of which 5 submitted their final results. We introduced a baseline system to benchmark performance in this area. This paper provides a comprehensive overview of the NCCE task, detailing the task’s framework, dataset characteristics, evaluation protocols, and metrics, alongside a summary of the submission results. We anticipate that the task will significantly aid in corpus construction and enhance the semantic understanding of the Chinese language.