A new discussion paper critically evaluates SemEval-2020 Task 1, a prominent benchmark for detecting lexical semantic change. The paper argues that the benchmark's operationalization is too narrow, focusing primarily on discrete sense changes rather than broader linguistic shifts. Furthermore, it highlights significant data quality issues, including corpus noise and preprocessing errors, which can distort results and hinder reproducibility. The authors call for future datasets and tasks to adopt more comprehensive theories of semantic change, transparent preprocessing documentation, and broader linguistic coverage to ensure more valid and generalizable progress in the field. AI
RANK_REASON This is a discussion paper evaluating a specific academic benchmark. [lever_c_demoted from research: ic=1 ai=0.7]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
