SHA, ALYSSA SHUANG ; Nunes, Bernardo Pereira ; HALLER, ARMIN . Link Topics from Q&A Platforms using Wikidata: A Tool for Cross-platform Hierarchical Classification. In: WebSci ’23: 15th ACM Web Science Conference 2023, 2023, Austin TX USA. Proceedings of the 15th ACM Web Science Conference 2023. New York: ACM, 2023. p. 357-362. doi: 10.1145/3578503.3583625
Link Topics from Q&A Platforms using Wikidata: A Tool for Cross-platform Hierarchical Classification
Authors
Alyssa Shuang Sha (ANU)
Bernardo Pereira Nunes (ANU)
Armin Haller (ANU)
Abstract
This paper proposes a novel rule-based topic classification tool for questions on Q&A platforms mediated by the Wikidata ontology – an open and accessible multilingual ontology curated by a large community of online users. Q&A platforms are important sources of information on the Web and often appear as part of Web search results. By adopting Wikidata taxonomic relations as references, our tool can categories the Web content from different platforms in a unified coarse-to-fine mode based on their domain coverage. To validate and demonstrate the potential applicability of our tool, a set of use cases and experiments are carried out on two popular Q&A platforms – Zhihu and Quora, where the impact of topic categories on question lifecycles is explored. Furthermore, we compare our results with the output generated by GPT-3 classifier. This tool sheds light on how structured knowledge bases can enable data interoperability and serve as a filtering functionality to mitigate classification bias of OpenAI.
Keywords:
Topic classification, Wikidata ontology, Entity Linking, Q&A platforms
