Recent Innovations: Short Papers and Posters

This concept paper focuses on what role metadata play in an AI lifecycle and how metadata research can ride out this AI wave with innovative creations. Specifically, we explore metadata’s role and potential related to data quality and ML models. The multidimensionality of metadata for data in AI is driving metadata to be micro-specific, embedded in data and models, highly computational, and fast-moving or agile. While there are no universally agreeable metadata schemas for documenting the artifacts in ML model development, there are some common areas or types of metadata for ML models. Data quality and ML models are tightly connected and can impact one another in significant ways. Trustworthy AI must rely on quality data and responsible, ethical, reproducible, verifiable ML models, and the assurance of these data and ML model properties relies on metadata. The complex, fast paced, and highly computational nature of metadata for AI artifacts (datasets, models, pipelines, algorithms, lineages, etc.) is making conventional metadata development processes and methods outdated, but meanwhile has prompted some innovative metadata creations.

The international BIBFRAME Interoperability Group (BIG) was initiated by the Program for Cooperative Cataloging (PCC) with the goal to "work collaboratively on the development and maintenance of interoperable BIBFRAME data guidelines to support production level implementation, to address issues restricting interoperability, and to inform development of associated toolings and infrastructure." (BIBFRAME Interoperability Group charge, April 15, 2022). The idea originated at the PCC BIBFRAME Data Exchange Meeting (September 2021), where the participants identified different implementation decisions of the BIBFRAME ontology as major obstacles to successful BIBFRAME data exchange. The conclusion was reached that an international group was needed to continue the conversation in this area.

BIG officially took up its work in June 2022. In its first year of existence, BIG reviewed the output of several related working groups, conducted a BIBFRAME implementation survey and analyzed the results, helped organize the 2022 Linked Data Summit held at the Library of Congress in November 2022, and developed a work plan in response to the recommendations from the Summit.

This work plan outlines four working areas: 1) defining a standard BIBFRAME “shape” necessary for data exchange utilizing PCC data and standards as a starting point; 2) creating recommendations that are readable by both technical staff and librarians (preferably only necessitating updates to be made in one place); 3) codifying the interoperability scope; and 4) documenting best practices for technical aspects of BIBFRAME interchange as identified through the work of the group and sharing with consultants for testing and validation of assumptions.

This poster presentation will first introduce the BIBFRAME Interoperability Group, including its charge, governance, and membership as well as the process for assembling this group. It will then go over the work plan, focusing on sharing our experience with using DC Tabular Application Profiles (DC TAP) and generating SHACL representation of the application profile. Lastly, the presentation will outline the next steps that BIG is planning to take, including determining patterns from concrete use cases to anchor what practical value the various profile decisions stem from.

This paper presents the implementation of a Linked Data Management System (LDMS) at the National Library Board Singapore (NLB), which aims to provide a unified view of bibliographic descriptions from diverse collections across the National Library, the Public Libraries, and the National Archives. The National Library Board worked with a vendor to convert metadata from multiple sources into entities in a triplestore for use in resource discovery. This paper will outline the challenges faced and lessons learnt in the production of Linked Data and improving data quality. The global pandemic forced the team to work remotely with an overseas vendor, which compounded the complexity of communicating relatively complex concepts and troubleshooting data issues. Great emphasis was placed on determining causes for data issues and correcting these. Challenges arose from the insertion of URIs into the source records to identify entities matching the string label as the reconciliation of entities extracted from different source systems is based on similarity of the name label and associated properties. The paper concludes with an outlook on the continuous refinement of data quality and the development of public interfaces to demonstrate the benefits of Linked Data to stakeholders. The development of this service inserts the discovery aspect directly into resources, demonstrating the potential of Linked Data to shape future services.