Three new UKRI projects strengthen KMi’s leadership in AI for Metascience 👉 blog.stem.open.ac.uk/three-new-uk... @fosborne.bsky.social @angelosalatino.bsky.social #AI #Metascience #ResearchAssessment #UKRI #KnowledgeGraphs #TrustworthyAI

Great news! Our team has secured three new UKRI metascience projects worth approximately £500,000 in total! Helping shape how science is evaluated and funded. blog.stem.open.ac.uk/three-new-uk... #Metascience

First results from our UKRI metascience project: one of the first studies of LLM grant scoring, using anonymised data within UKRI's secure environment. Insights for AI-assisted grant review. More to follow! arxiv.org/abs/2609.25327 #Metascience #PeerReview

🎉 Our SEER paper is now published in Knowledge-Based Systems! SEER combines AI agents to extract who said what in the news and build knowledge graphs, supporting analysis of voices and viewpoints at scale. doi.org/10.1016/j.kn... #AI #KnowledgeGraphs

🎉 Our CSO Classifier 4.0 paper is now published in SoftwareX! The open-source tool maps research papers to Computer Science Ontology topics, with an updated semantic model, topic weights and faster processing. doi.org/10.1016/j.so... #OpenScience #CSOClassifier

AI can help researchers identify plausible directions for what to study next. Our paper presents ResearchLink, which generates and ranks hypotheses from scientific knowledge graphs for human investigation. Paper: doi.org/10.1016/j.kn... #AIForScience #KnowledgeGraphs

AI systems need data platforms that can evolve with them. Our Journal of Big Data paper presents an open, knowledge-graph-based framework for modelling adaptable platforms. Paper: doi.org/10.1186/s405... #KnowledgeGraphs #BigData

Knowledge graphs need scalable validation as automated generation improves. Our paper shows how LLMs, automated checks and targeted expert review can share the work. Paper: doi.org/10.1016/j.ip... #KnowledgeGraphs #HumanInTheLoop

Science needs maps that evolve as quickly as research itself. In our paper, we show that LLMs, including compact open models, can identify the topic relationships needed to build scholarly ontologies. Read the paper: doi.org/10.1016/j.ip... #AIinScience #Ontologies

Could AI assess research without becoming another black box? Our preprint identifies interpretable signals that LLMs associate with scientific quality, from methods to specialised language. Read the paper: arxiv.org/abs/2602.19115 #AIinScience #XAI

How Do LLMs Encode Scientific Quality? An Empirical Study Using Monosemantic Features from Sparse Autoencoders

In recent years, there has been a growing use of generative AI, and large language models (LLMs) in particular, to support both the assessment and generation of scientific work. Although some studies ...

arxiv.org

Research teams benefit from combining different expertise. Our Scientometrics paper links expertise diversity to five-year citation impact across 944,508 Computer Science papers. Read the paper: doi.org/10.1007/s111... #ScienceOfScience #TeamScience

A useful literature review shows what is known, unresolved and how studies change the picture. Our new PeerJ Computer Science paper is out: Sci-Sentence helps AI recognise these roles at scale. Read the paper: doi.org/10.7717/peer... #AIinScience #NLP

Tanay Aggarwal, Angelo Salatino, Francesco Osborne, Enrico Motta: Benchmarking Resource-Efficient LLMs for Research Topic Ontology Generation in the Biomedical Field https://arxiv.org/abs/2607.17902 https://arxiv.org/pdf/2607.17902 https://arxiv.org/html/2607.17902

AI could help build connected maps of science across disciplines. Our new ACM TIST paper is out: multidisciplinary fine-tuning transfers strongly across fields, supporting better research search and discovery. Read the paper: doi.org/10.1145/3828... #AIinScience #Ontologies

Leveraging Large Language Models for Generating Research Topic Ontologies: A Multi-Disciplinary Study | ACM Transactions on Intelligent Systems and Technology

Ontologies and taxonomies of research fields are critical for managing and organising scientific knowledge, as they facilitate efficient classification, dissemination and retrieval of information. How...

doi.org

News analysis should capture viewpoints and voices, not only topics. Our new paper is out: we combine fine-tuned LLMs with Wikidata context for interpretable analysis of political debate. Read the paper: doi.org/10.1140/epjd... #NewsAnalytics #KnowledgeGraphs

Integrating Large Language Models and knowledge graphs to capture political viewpoints in news media - EPJ Data Science

News sources play a central role in democratic societies by shaping political and social discourse through specific topics, viewpoints and voices. Understanding these dynamics is essential for assessing whether the media landscape offers a balanced and fair account of public debate. In earlier work, we introduced a pipeline that, given a news corpus, i) uses a hybrid human–machine approach to identify the range of viewpoints expressed about a given topic, and ii) classifies relevant claims with respect to the identified viewpoints, defined as sets of semantically and ideologically congruent claims (e.g., positions arguing that immigration positively impacts the UK economy). In this paper, we improve this pipeline by i) fine-tuning Large Language Models (LLMs) for viewpoint classification and ii) enriching claim representations with semantic descriptions of relevant actors drawn from Wikidata. We evaluate our approach against alternative solutions on a benchmark centred on the UK immigration debate. Results show that while both mechanisms independently improve classification performance, their integration yields the best results, particularly when using LLMs capable of processing long inputs.

doi.org

Online discussions can influence markets, but sentiment alone misses what people are actually recommending. We introduce Social Trading Action Detection to identify buy, sell or no-action signals. doi.org/10.1016/j.ip... #FinancialNLP #LLMs

Can small open LLMs help build biomedical ontologies? We tested five models on MeSH relations. Targeted fine-tuning produced strong results, even for a 3B-parameter model. Paper, benchmark and code: arxiv.org/abs/2607.17902 #BiomedicalAI #Ontologies

Benchmarking Resource-Efficient LLMs for Research Topic Ontology Generation in the Biomedical Field

Knowledge Organization Systems like Ontologies and taxonomies are fundamental for structuring scientific knowledge, yet their manual curation presents a persistent bottleneck in knowledge management. ...

arxiv.org

🚀 Can AI generate research hypotheses? ResearchLink does: a domain-independent method that predicts novel links over scientific KGs, outperforming strong baselines with excellent precision across multiple domains. https://doi.org/10.1016/j.knosys.2025.113280 #AIinScience #KnowledgeGraphs

🔬 AI can analyse news at scale, but first we need to define what a story can be *about*. Our Semantic Web Journal paper proposes a rigorous typology of news topics, realised as reusable ontologies (50 classes, 1,622 axioms). https://doi.org/10.1177/22104968251344461 #SemanticWeb #AI

🔥 AI is transforming scientific discovery, but we need structured knowledge to reason over it. CS-KG 2.0 captures 25M entities and 67M relationships from 15M CS papers. Open via SPARQL. New in Scientific Data (Nature). https://doi.org/10.1038/s41597-025-05200-8 #KnowledgeGraphs #OpenScience

📣📣 Researchers and practitioners in NLP, Semantics, Knowledge Engineering, AI, Neurosymbolic AI: this is a call for you! We have various tracks. We look forward to receiving your manuscript

SEMANTiCS Conference@semantics-conf.bsky.social · 8mo ago

🔔 2nd JOINT CALL FOR SUBMISSIONS #SEMANTiCS 2026 invites original contributions exploring the latest advances in Linked Data and semantic-enabled AI systems. Important dates: 2026-eu.semantics.cc/page/importa... ▶️ Help us spread the word!

🔬 Stronger science needs teams that combine complementary expertise. Our Scientometrics study of 944K CS papers shows diversity of expertise is a major driver of citation impact (r=0.955). https://doi.org/10.1007/s11192-026-05560-x #ScienceOfScience #Scientometrics

Can LLMs build scholarly ontologies? We tested 17 models and the results are striking: top F1-score of 0.967, and even a small 7B model hit 0.92. New paper in Information Processing & Management: https://doi.org/10.1016/j.ipm.2025.104262 #AI #KnowledgeGraphs

🔥 Generative AI is reshaping scientific research. But do these systems actually understand science? In our new paper, we peek inside LLMs to understand how they encode scientific quality, revealing several monosemantic features that predict research quality. 📄 arxiv.org/pdf/2602.19115

🔥 Generative AI can now answer complex scientific questions! Our latest paper evaluates the ability of #LLMs to answer questions over #KnowledgeGraphs through Natural Language to #SPARQL translation. Excellent results across multiple benchmarks. 📄 dl.acm.org/doi/10.1145/...

🔥 Generative #AI is revolutionising corporate sustainability assessment! In our last paper, we optimised AI agents to identify #ESG activities in corporate documents, boosting transparency and compliance with EU #environmental regulations. 📄https://dl.acm.org/doi/full/10.1145/3768292.3770371

📢 Call for Submissions Open! Join us for #SEMANTiCS2026. Let's discuss the future of #LinkedData & #Semantic #AI! 📍Ghent, BE 📅 Sept 15-17, 2026 🔹 Research & Innovation 🔹 Industry & Use Case 🔹 Posters & Demos 🔹 Workshops & Tutorials Calls: 2026-eu.semantics.cc Please, share! #SemanticsConf

KMi and Springer Nature renew their partnership for another two years kmi.open.ac.uk/news/article... #AIInnovation #ScholarlyPublishing #KnowledgeMining #ResearchIntelligence #AcademicIndustryPartnership #DigitalTransformation #Scientometrics #SpringerNature

Our survey paper on #AI tools for literature reviews has been referenced by the new EU report on the Role of AI in Research! We analyse 32 groundbreaking tools and discuss key challenges. #LLMs are revolutionising the field! 📄 link.springer.com/article/10.1... #DataScience #ML #Research

Generative #AI is a powerful ally in advancing the #SustainableDevelopmentGoals. Our latest paper shows that small open-source #LLMs can be optimised to accurately identify #SDG content, paving the way for a new generation of monitoring tools. 📄 doi.org/10.1109/ACCE... #ML #Sustainability #Research