Expert Finder Systems International Forum 2025

Carnegie Mellon University Libraries is currently developing an innovative suite of AI-powered tools addressing two critical challenges in research information management: the adoption barrier for public researcher profiles and the balanced distribution of committee service responsibilities among faculty. Our prototype toolset leverages natural language processing and machine learning techniques to extract committee service data from faculty CVs and other unstructured sources, import this information into the Elements/Scholars@CMU platform, and analyze the data to provide recommendations for achieving greater balance in committee service workload. This presentation will share our in-progress development approach, initial prototype functionality, methodological considerations, and preliminary findings. We’ll demonstrate how libraries can position themselves as essential curators of institutional data while advancing fairness initiatives through innovative applications of AI in research information management.

Our university’s Research Information Management (RIM) system, Elements, was rolled out for Faculty Activity Reports (FARs) in 2015, initially without formal requirements. By 2022, faculty from all colleges were required to manage their activity data within Elements, which integrates with a homegrown system for FARs. While data is set to internal by default, faculty can choose to share data from Elements in our public-facing profile system. Given the parallel evolution of systems and internal requirements over time, our Team recognized the need to evaluate our current systems and communications to best meet stakeholder needs.

This presentation focuses on analysis of existing data in our Elements system, such as: user activities and profile completeness, with consideration of user subgroups. We analyzed reported activities: distribution of scholarly work types, number and type of manually created records, error corrections, and updates. We investigated user decisions, such as how they claim scholarly identifiers like ORCID, if they share a profile photo, when they deposit works to the institutional repository, and if they share data publicly. The analysis aims to identify factors influencing RIM utilization, such as college, faculty rank, gender; and to highlight areas for improvement in user engagement and data accuracy.

By understanding how faculty use the RIM system, we can make targeted system improvements. This includes refining FAR template designs and system enhancements. These improvements can lead to more efficient data entry and increased faculty satisfaction. Improving data quality will enhance the impact of the system, enabling more accurate insights about and evaluation of faculty activities and contributions. The findings will inform potential next steps, such as surveys, focus groups, and user experience studies of specific system functions, features, or use cases, further enhancing the system’s value to stakeholders.

Universities increasingly recognize the value of interdisciplinary collaboration as a powerful way to tackle complex challenges. However, establishing effective partnerships across institutions, industry, and government is often difficult due to siloed organizational structures and language barriers between sectors.

Simply bringing potential collaborators together is often not enough: effective partnerships depend on clear communication and shared understanding. Industry practitioners often describe emerging technologies in terms unfamiliar to academics, who instead reference more traditional disciplinary frameworks. This linguistic disconnect creates challenges when industry users seek academic expertise.

To address these dual challenges of language barriers and researcher time constraints, OIEx is partnering with Digital Science to explore the potential of two complementary AI-powered approaches:

• Generative AI-driven search: Integrating generative AI models within researcher profile searches to translate and interpret user queries from industry partners, allowing for intuitive discovery of academic expertise despite differing vocabularies and linguistic approaches.
• Automated tagging of researcher profiles: An exploratory prototype is underway to automatically enhance academic researcher profiles with industry-specific keywords and standardized industry classifications, such as the North American Industry Classification System (NAICS).