Measuring Fine-Grained Domain Relevance of Terms: A Hierarchical Core-Fringe Approach

Abstract

We propose to measure fine-grained domain relevancethe degree that a term is relevant to a broad (e.g., computer science) or narrow (e.g., deep learning) domain. Such measurement is crucial for many downstream tasks in natural language processing. To handle longtail terms, we build a core-anchored semantic graph, which uses core terms with rich description information to bridge the vast remaining fringe terms semantically. To support a finegrained domain without relying on a matching corpus for supervision, we develop hierarchical core-fringe learning, which learns core and fringe terms jointly in a semi-supervised manner contextualized in the hierarchy of the domain. To reduce expensive human efforts, we employ automatic annotation and hierarchical positive-unlabeled learning. Our approach applies to big or small domains, covers head or tail terms, and requires little human effort. Extensive experiments demonstrate that our methods outperform strong baselines and even surpass professional human performance. 1

Extraction and Association Analysis of Risk Factors in Gas Pipeline Network Emergencies Through Fusing Expert Knowledge and Pre-Trained Model

“…The initial extraction of risk factors often includes a lot of irrelevant information. Applying domain relevance, a filtering technique based on domain-specific knowledge, significantly improves the analysis's precision and efficiency [25]. By evaluating the domain significance of identified risk factors, this study efficiently filters out unrelated information.…”

Domain Representative Keywords Selection: A Probabilistic Approach

“…However, the main focus of the above studies is to describe single/multiple documents rather than domain-specific keywords extraction. To solve this problem, several researches (Liu et al, 2015;Shang et al, 2018;Lu et al, 2019;Wang et al, 2020;Huang et al, 2021) have been conducted on domain-specific fine-grained keyword extraction. E.g., (Huang et al, 2021) propose an algorithm for measuring the relevance of a keyword in a particular domain.…”

“…To solve this problem, several researches (Liu et al, 2015;Shang et al, 2018;Lu et al, 2019;Wang et al, 2020;Huang et al, 2021) have been conducted on domain-specific fine-grained keyword extraction. E.g., (Huang et al, 2021) propose an algorithm for measuring the relevance of a keyword in a particular domain. However, this approach requires a user to provide some seed domain-relevant terms for supervising the algorithm.…”

“…However, their main focus is to extract terms from single/multiple documents without considering whether the extracted terms are distinctive to a target domain contrastive with a context. There is also some previous research (Liu et al, 2015;Shang et al, 2018;Lu et al, 2019;Huang et al, 2021) that tries to find fine-grained domain-specific keywords from the text. However, these approaches mostly rank keywords based on their specificity to a corpus (or domain) rather than selecting a predefined number of keywords with a global objective of representing the target domain.…”