Search

This paper describes our experience in combining a large corpus of material that was classified manually over decades of Art His- torical and Book History research using the ICONCLASS4 subject clas- sification system with heuristics based on a current state-of-the-art neu- ral network (CLIP from OpenAI), leveraging visual similarity to provide suggestions for automated classification of cultural heritage content. The effectiveness of the approach is demonstrated through an evaluation of the underlying extreme multi-label classification problem.

Ontologies contain an abundance of concepts, are frequently structured as hierarchies, and can cover different domains of knowledge. Polysemous concepts need to be disambiguated for annotation purposes, for example, a concept such as depression has a different meaning in the fields of psychology and economics. In this paper, we introduce the use of the maximum mean discrepancy to indicate whether sets of concepts sharing the same meaning should be merged. This method is a novel approach to ontology maintenance because it provides an objective metric that supports the decision-making of subject matter experts during the concept evaluation process. Our objective is thus to assist ontology maintenance, in particular the organization of concepts, through an analysis framework that gives insights into the polysemy of concepts.

There were 7 papers submitted for peer-review to this workshop. Out of these, 4 papers were accepted for this volume.

Knowledge graphs in both public and corporate settings need to keep pace with the constantly growing amount of data being generated. It is, therefore, crucial to have automated solutions for assessing the quality of Knowledge Graphs, as manual curation quickly reaches its limits. This research proposes the use of KG-BERT for a triple (binary) classification task that assesses the quality of a Knowledge Graphs’s hierarchical structure. The use of KG-BERT allows the textual as well structural aspects of a Knowledge Graph to be leverage for this quality assessment (QA) task. The performance of our proposed approach is measured using four different Knowledge Graphs: two branches (Physics and Mathematics) of a corporate Knowledge Graph - OmniScience, a WordNet subset, and the UMLS Semantic Network. Our method yields high-performance scores on all four KGs (88-92% accuracy) making it a relevant tool for quality assessment and knowledge graph maintenance.

Materials are either enabler or bottleneck for the vast majority of technological innovations. The digitization of materials and processes is mandatory to create live production environments which represent physical entities and their aggregations and thus allow to represent, share, and understand materials changes. However, a common standard formalization for materials knowledge in the form of taxonomies, ontologies, or knowledge graphs has not been achieved yet. This paper sketches the efforts in modelling an ontology prototype to describe Materials Science experiments. It describes what is expected from the ontology by introducing a use case where a process chain driven by the ontology enables the curation and understanding of experiments.

Exponential computational cost arises when graph convolutions are performed on large graphs such as knowledge graphs. This computational bottleneck, dubbed the ‘neighbor explosion’ problem, has been overcome through application of graph sampling strategies. Graph Convolutional Network architectures that employ such a strategy, e.g. GraphSAGE, GraphSAINT, circumvent this bottleneck by sampling subgraphs. This approach improves scalability and speed at the cost of information loss of the overall graph topology. To improve topological information retention and utilization in graph sampling frameworks, we introduce Graph Position-aware Preprocessed Embeddings (GraphPOPE), a novel, feature-enhancing preprocessing technique. GraphPOPE samples influential anchor nodes in the graph based on centrality measures and subsequently generates normalized geodesic, Cosine or Euclidean distance embeddings for all nodes with respect to these anchor nodes. Structural graph information is retained during sampling as the position-aware node embeddings act as a skeleton for the graph. Our algorithm outperforms GraphSAGE on a Flickr benchmark dataset. Moreover, we demonstrate the added value of topological information to Graph Neural Networks.

This poster focuses on capturing the temporal evolution of migration-related topics on relevant tweets. It uses Dynamic Embedded Topic Model (DETM) as a learning algorithm to perform a quantitative and qualitative analysis of these emerging topics. TweetsKB is extended with the extracted Twitter dataset along with the results of DETM which considers temporality. These results are then further analyzed and visualized. It reveals that the trajectories of the migration-related topics are in agreement with historical events.