Tzu-Kun Hsiao's Dissertation Defense

Citation links are commonly used as proxies for knowledge diffusion in the scientific literature because they connect past and present research. However, citation links alone provide limited information about how prior research is described, discussed, or used in later studies. Citation contexts can enrich these past-to-present connections with semantic information because they convey citing authors’ interpretations of the cited work. By investigating citation contexts, we can gain insights into the reasons behind citations (i.e., citation purposes), the opinions of citing authors toward the cited articles (i.e., citation sentiments), and the specific knowledge being referenced (e.g., findings or methods). This dissertation begins by addressing the challenge of obtaining high-quality citation context data at scale and demonstrates how the semantics associated with citations can supplement analyses of knowledge diffusion in the scientific literature. In specific, this dissertation has three thrusts: The first thrust addresses the practical but non-trivial issues of identifying citation contexts from a large collection of scientific articles. The second thrust investigates the semantics of citation contexts mentioning retracted articles using content analysis and natural language processing (NLP) techniques. The third thrust presents an analysis of citation contexts from the perspective of semantic diversity using pre-trained sentence embeddings.

Questions? Contact Tzu-Kun Hsiao.