RT @Cardiff_NLP: 🚨 Check out the Cardiff NLP group and collaborators' accepted papers at #ACL2025NLP (Main, Findings, Demo, Workshops) and….

I am looking for a postdoctoral research associate to work on (LLM-based and neurosymbolic) reasoning for story understanding, as part of the EPSRC-funded ReStoRe project. Details: (deadline 21st July).@Cardiff_NLP @cardiff_krr.

This is a joint project with @diofantino (Cardiff School of Chemistry) and Ondrej Kuzelka (CTU Prague), where we will be looking at applications in computational chemistry, natural language understanding, and knowledge graph reasoning.

I'm looking for a postdoctoral research associate to work on InteGraL (“Interpretable Graph-Based Machine Learning”), a three year project funded by the Leverhulme Trust. Deadline: 4th June.Details: @cardiff_krr @Cardiff_NLP.

One week left to apply for this postdoc position in Cardiff!.

RT @ArtemShelmanov: The Call for Volunteers for COLING 2025 is now open! Apply by December 6th, AoE. Accepted volun….

I'm looking for a postdoc to work on document-level natural language understanding and neurosymbolic reasoning. Deadline for applications: 25 November.Full details: @Cardiff_NLP @cardiff_krr.

I am looking for a postdoc to work on neurosymbolic methods (broadly interpreted) for natural language understanding. Available from: 1st November.Deadline for applications: 7th October.More details: @Cardiff_NLP @cardiff_krr.

RT @coling2025: #NLProc #coling2025.📣 Final Call for Papers!. COLING 2025 invites the submission for main conference papers. Deadline for….

The deadline for this position is in just a few days .

We're presenting two posters at #acl2024 tomorrow:. Ranking Entities along Conceptual Space Dimensions with LLMs: An Analysis of Fine-Tuning Strategies: Modelling Commonsense Commonalities with Multi-Facet Concept Embeddings:

Going from hexagons to octagons allows us to capture arbitrary knowledge graphs. Moreover, the resulting relation embeddings are then closed under both intersection and composition, unlike for previous region based embeddings.

Adding domain and range constraints leads to a model based on hexagons, but this model is still not fully expressive.

For instance, TransE can be recast as a region based approach, where the geometric regions correspond to “bands”

The main advantage KG embedding models of geometric regions is that we can capture rules, in addition to the facts of the KG. This makes it possible to impose constraints on the model or to inject prior knowledge among others.

We have a poster on “Capturing Knowledge Graphs and Rules with Octagon Embeddings” at #IJCAI2014 tomorrow (Thursday) at session 3F-Halla B. Paper: @cardiff_krr @Cardiff_NLP.

At the methodological level, some of the core challenges include: the need for dealing with noisy inputs, the need for systematic generalisation, and the need for learning inference patterns from few observations. Deadline for applications: 1st September.

The overall aim of the project is to develop commonsense reasoning strategies that allow us to go beyond traditional information extraction, converting text into high quality knowledge graphs that capture both what is explicitly stated and what can be plausibly inferred.

I am looking for a postdoctoral research associate to work on neurosymbolic methods for story understanding, as part of an EPSRC project on “Reasoning about Structured Story Representations”. Details: @Cardiff_NLP @cardiff_krr.#NLProc #ML #neurosymbolic.