Keynotes

Knowledge Beyond Documents: FAIR Ontologies for Humans and Machines

The FAIR principles have significantly improved the publication and reuse of digital resources, promoting data that are Findable, Accessible, Interoperable, and Reusable. However, as AI-based systems increasingly consume and generate knowledge, achieving machine-actionable interoperability requires not only FAIR data but also FAIR ontologies. In this keynote, I will argue that FAIR ontologies are a cornerstone of machine-actionable knowledge. Beyond enabling semantic interoperability, they support knowledge integration, discovery, and explainability across digital infrastructures. I will present recent advances in ontology engineering, discussing practical approaches, tools, and best practices for designing and publishing FAIR ontologies, together with current challenges such as documentation, metadata, versioning, quality assessment, governance, modularity, reuse, and long-term sustainability. Ultimately, I will argue that moving from FAIR Data to FAIR Knowledge requires treating ontologies as first-class FAIR digital resources.

Temporal Intelligence in Language Technologies: Modeling, Retrieval, and Reasoning

Time is central to how we understand the world. It shapes how we interpret events, construct narratives, and reason about change, causality, and evidence. In this talk, I will first examine how well large language models handle temporal information and temporal reasoning, an essential but still less explored dimension of language intelligence. I will then discuss recent advances in incorporating temporal awareness into language technologies, including time-aware representation learning, retrieval from evolving document collections, and the development of new resources for training and evaluation. I will also introduce temporal reasoning-oriented question answering and retrieval benchmarks that highlight the challenges of retrieving time-sensitive information, answering complex temporal questions, and reasoning over changing evidence. The talk will conclude with open challenges for building language technologies that can better understand, retrieve, and reason over information across time.

Transitive Credit Allocation for Heterogeneous and Dynamic Scholarly Graphs

Citation counts and the h-index stop at the first link, so the methods, datasets, and software behind a cited work receive nothing from the citation. Scholarly graphs now link papers to datasets, software, clinical trials, and patents through typed relations, and versioning makes these graphs dynamic and can introduce cycles.
This talk presents Transitive Credit Allocation (TCA). TCA is joint work with Peter Buneman, Mirco Cazzaro, Matteo Lissandrini, and Laura Menotti. Each citation brings one unit of credit to the cited work. Each work keeps a share of the credit it receives, its retention, and passes the rest equally to the works it cites. The model conserves credit in citation units, returns citation counts at 100% retention, and sends credit further back as retention decreases. The same rule applies to datasets, software, and authorized versions, and credit circulating in a cycle settles because every work keeps a positive share.
On a citation graph of about 40 million papers, patents, and clinical trials from the PubMed Knowledge Graph, we computed author h-indices at four retention levels from the credit each paper keeps. At 25% retention, 91% of the authors in the top 500,000 under both citation counting and TCA see their h-index fall, while 119 gain 50 points or more. Foundational authors climb, and a 1955 paper with 19 citations receives 5,448 units of credit from later work.
The talk closes with open questions on setting the retention, on the quality of data citations and scholarly graphs, and on crediting the data behind LLM answers.