Cold-start knowledge base population is concerned with populating an empty knowledge base from scratch. In this paper, we describe a method that supports cold-start knowledge base population by information extraction from unstructuredtext. The method is based on a given ontology that defines the main classes and properties for the domain in question. The information extraction approach capitalizes onprobabilistic graphical models, in particular factor graphs. In this paper, we describe (i) the probabilistic model underlyingour approach, (ii) the inference techniques that are based on Markov Chain Monte Carlo sampling over possible schema instantiations defined by the ontology and conditioned on a given text, as well as (iii) the approach to learn the parameters of the model. Many other methods such as open information extraction rely on extracting single binary relations using classification methods. In contrast, our method takes rigorous account of an ontology and extracts more complex ontology-compliant structures that capture the meaning of a text in a more comprehensiveway. We present results on two datasets: a novel dataset of scientific publications covering pre-clinical studies in the spinal cord injury domain, and the well-known, but now quite aged, Message Understanding (MUC) dataset.

This is a quick survey about efficient search on a text corpus combined with a knowledge base. We provide a high-level description of two systems for searching such data efficiently. The first and older system, Broccoli, provides a very convenient UI that can be used without expert knowledge of the underlying data. The price is a limited query language. The second and newer system, QLever, provides an efficient query engine for SPARQL+Text, an extension of SPARQL to text search. As an outlook, we discuss the question of how to provide a system with the power of QLever and the convenience of Broccoli. Both Broccoli and QLever are also useful when only searching a knowledge base (without additional text).

We discuss some essential issues for the formal representation of norms to implement normative reasoning, and we show how to capture those requirements in a computationally oriented formalism, Defeasible Deontic Logic, and we provide the description of this logic, and we illustrate its use to model and reasoning with norms with the help of legal examples.

Large-scale probabilistic knowledge bases are becoming increasingly important in academia and industry alike. They are constantly extended with new data, powered by modern information extraction tools that associate probabilities with knowledge base facts. This tutorial is dedicated to give an understanding of various query answering and reasoning tasks that can be used to exploit the full potential of probabilistic knowledge bases. In the first part of the tutorial, we focus on (tuple-independent) probabilistic databases as the simplest probabilistic data model. In the second part of the tutorial, we move on to richer representations where the probabilistic database is extended with ontological knowledge. For each part, we review some known data complexity results as well as discuss some recent results.

With proliferation of semantic data, there is a need to cope with trillions of triples by horizontally scaling data management in the cloud. To this end one needs to advance (i) strategies for data placement over compute and storage nodes, (ii) strategies for distributed query processing, and (iii) strategies for handling failure of compute and storage nodes. In this tutorial, we want to review challenges and how they have been addressed by research and development in the last 15 years.

Automated rule-based reasoning is a useful technique to extract new information from knowledge graphs (KGs). Unfortunately, it is challenging to apply it to real-world KGs (such as Wikidata or DBPedia) due to the large size of these datasets. In this lecture, I will first introduce which are the difficulties that modern reasoners must face in order to be applied on large inputs, and describe the most effective techniques which concern and efficient execution of safe and existential rules. Finally, I will discuss several state-of-the-art systems, analyse some experimental evaluations, and point out to interesting directions for future research.

Large-scale cross-domain knowledge graphs, such as DBpedia or Wikidata, are some of the most popular and widely used datasets of the Semantic Web. In the first part of this lecture, you will learn which knowledge graphs exist on the Semantic Web. I will introduce a few typical machine learning tasks, such as product recommendation and social media analysis, and discuss how knowledge graphs can be used to improve the performance within those tasks. In the second part of the lecture, we will discuss how machine learning can help improving existing knowledge graphs, tackling typical problems such as link and type prediction or error identification.

Advances in information extraction have enabled the automatic construction of large knowledge graphs (KGs) like DBpedia, Freebase, YAGO and Wikidata. Learning rules from KGs is a crucial task for KG completion, cleaning and curation. This tutorial presents state-of-the-art rule induction methods, recent advances, research opportunities as well as open challenges along this avenue. We put a particular emphasis on the problems of learning exception-enriched rules from highly biased and incomplete data. Finally, we discuss possible extensions of classical rule induction techniques to account for unstructured resources (e.g., text) along with the structured ones.

The goal of the lecture is to outline how to design, develop and deploy a stream processing application in a web environment. To this extent, the lecture (1) surveys existing research outcomes from the Stream Reasoning, (2) introduces RDF stream processing techniques as powerful tools to use when addressing a data-centric problem characterised both by variety and velocity (such as those typically found on the modern Web), (3) presents a relevant Web-centric use-case that requires to address simultaneously data velocity and variety, and (4) guides the participants through the development of a Web stream processing application.