2018 | OriginalPaper | Chapter
Mandarin Relata: A Dataset of Word Relations and Their Semantic Types
Authors : Hongchao Liu, Chu-Ren Huang, Ren-kui Hou
Published in: Chinese Lexical Semantics
Publisher: Springer International Publishing
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
For both the training and evaluation of semantic distributional models, language datasets are needed that are both elaborate in their word level descriptors and readily intuitive to human judgment. The current paper introduces a dataset for Mandarin Chinese constructed through the combination of word relation pairs from two distinct sources: corpus extraction, and human elicitation. Our results show that while more word relation pairs were gained through the corpus extraction process, human elicited semantic neighbors were almost twice as likely to show agreement with human raters. The current methods created 4091 word relation pairs that span hypernymy, hyponymy, synonymy, antonymy, and meronymy alongside semantic type information. To date, this is the largest collection of human-rated word relation pairs in Mandarin Chinese.