ChiWUG: A Graph-based Evaluation Dataset for Chinese Lexical Semantic Change Detection

Abstract

Recent studies suggested that language models are efficient tools for measuring lexical semantic change.In our paper, we present the compilation of the first graph-based evaluation dataset for semantic change in the context of the Chinese language, covering the periods before and after the Reform and Opening Up.Exploiting the existing framework DURel, we collect over 61,000 human semantic relatedness judgments for 40 targets.The inferred word usage graphs and semantic change scores provide a basis for visualization and evaluation of semantic change.

Authors 5

1. Jing Chen Aachen

   University of Stuttgart · Leiden University · Hong Kong Polytechnic University

   Affiliation as printed

   The Hong Kong Polytechnic University

   Jelena Prokic Leiden University

   University of Stuttgart

2. Emmanuele Chersoni Aachen

   University of Stuttgart · Leiden University · Hong Kong Polytechnic University

   Affiliation as printed

   The Hong Kong Polytechnic University

   Jelena Prokic Leiden University

   University of Stuttgart

3. Dominik Schlechtweg Aachen

   University of Stuttgart · Leiden University · Hong Kong Polytechnic University

   Affiliation as printed

   University of Stuttgart

   Jelena Prokic Leiden University

   The Hong Kong Polytechnic University

4. Jelena Prokić Aachen

   Leiden University

   Affiliation as printed

   Leiden University

5. Chu‐Ren Huang Aachen

   University of Stuttgart · Leiden University · Hong Kong Polytechnic University

   Affiliation as printed

   The Hong Kong Polytechnic University

   Jelena Prokic Leiden University

   University of Stuttgart