Explicit and Implicit Knowledge Extraction

EIKE

Co-located with FOIS 2025

Catania - Sicily

Beyond the Surface - Unraveling Hidden Knowledge in Text and Data.

Extracting nuanced and context-sensitive information (i.e., the subtle, often implicit data embedded in text, images, and multimodal signals) is a key challenge in advancing Knowedge Extraction (KE) for Entity Linking (EL), Information Retrieval (IR), and any sense-making application. While effective for explicit knowledge extraction, traditional pipelines often struggle to capture more complex elements, such as emotional undertones, sociocultural themes, or context-dependent subtleties. Recent advancements in machine learning, particularly Large Language Models (LLMs), show promise for directly inferring enriched semantic graphs that bridge this gap. The first edition of EIKE - Explicit and Implicit Knowledge Extraction - offers an opportunity to discuss breakthrough techniques, including neuro-symbolic systems, deep learning models, and ontology-based methods to address those challenges. By focusing on the direct extraction and representation of knowledge, the workshop aims to advance the state-of-the-art in semantically rich and context-aware knowledge systems.

This workshop accepts contributions on several topics, such as (but not limited to): knowledge extraction, information extraction, machine reading, ontology-driven contextualization, neuro-symbolic AI for EL and IR, and ethical aspects of knowledge extraction.

Topic Areas

Knowledge Extraction and Information Extraction: Techniques for identifying and representing explicit and tacit context-dependent knowledge embedded in unstructured and multimodal data sources, with an emphasis on socially relevant insights.
Machine reading and multimodal machine understanding: Exploring models and methodologies that employ a graph representation to represent knowledge from text and multimodal sources.
Ontology-Driven contextualization: Integrating ontologies to ground implicit and explicit knowledge in structured and interpretable frameworks.
Neural, symbolic and neuro-symbolic approaches for EL and IR: Neural, symbolic and combined approaches for richer, more accurate entity linking and information retrieval.
Semantics of implicit knowledge systems: What truth and modal operators underlie implicit knowledge axioms in neuro-symbolic systems?
Ethics of implicit knowledge systems: Addressing the transparency, biases, and societal implications of extracting and operationalizing social and human-focused knowledge.
Practical applications: Case studies in domains such as healthcare, legal systems, and cultural heritage that highlight the benefits of knowledge extraction for socially attuned applications.