Data-Driven Scholarly Intelligence: A Survey and Outlook

摘要

Digital scholarly platforms are reshaping scholarly communication by generating heterogeneous data beyond traditional publication and citation records. This survey reviews data-driven scholarly intelligence as a framework for organizing such data and transforming them into human-centered academic and knowledge services. We first characterize the scholarly data ecosystem and discuss how heterogeneous data can be structured through data organization and scholarly knowledge modeling. We then synthesize representative methods from graph-based modeling, recommendation, language understanding, knowledge grounding, and large-language-model-based agents. The survey further examines how scholarly intelligence can support academic interaction, knowledge matching, decision-oriented services, and research workflow assistance. Finally, we identify challenges related to data governance, fairness, explainability, evaluation, reproducibility, and trustworthy service design. By connecting scholarly data mining, academic social network analysis, and AI- enabled research services, this survey provides a compact view of how scholarly data can support evidence-grounded, explainable, and responsible scholarly intelligence.