Corpus Construction for Generating Knowledge Graph of Sichuan Cuisine

Xia Yang, Siyuan Jing, Xingyuan Chen, Peng Jin

Abstract: Sichuan cuisine and its culture play an important role in Chinese food culture. Unfortunately, people can only learn Sichuan cuisine through traditional ways such as the Web and books. To address the problem, we developed the annotation system of Sichuan cuisine and designed the annotation schemes of named entities and entity relations under the guidance of Sichuan cuisine masters. There are 220,000 words in the corpus that was annotated and proofread manually by an annotation system. We constructed a corpus of named entities and entity relations of Sichuan cuisine. It contains 691 kinds of Sichuan cuisine recipes, including 8,683 entities and 8,700 entity relations. The consistency of multiple-around annotation is 0.94 and 0.93, respectively. Based on the corpus, a knowledge graph of Sichuan cuisine was constructed and will be used to develop the tourism QA system in the future.

External IDs:dblp:conf/clsw/YangJCJ22