In distributed RDF stores the strategy how data is distributed over several compute nodes affects query performance. When hash-based data distribution strategies are used, the query workload tends to be equally balanced among all compute nodes while a relatively high number of intermediate results must be transferred between compute nodes. Graph-clustering-based approaches reduce the number of transferred intermediate results while the query workload becomes more imbalanced. This paper presents a novel data distribution strategy that combines the advantages of both strategies. To this end, we collocate the individuals of small sets of closely connected data items on compute nodes.
RDF offers a flexibility that is desireable in many situation such as data exchange, integration or knowledge representation. However, a program usually consumes precise substructures of such graph data. The
flexibility of RDF forces a developer to take special care when designing such programs in order to avoid runtime errors. Ideally, we could leverage type systems to avoid possible runtime errors. The Shape Constraint Language (SHACL) is a relatively new way of structural validation on RDF data sources. An Integration of SHACL into a type system can be used for an automatic detection of runtime errors.
To work with distributed ledger technologies such as blockchains, we need to understand the advantages and disadvantages of the different approaches. In this presentation we will look at three different concepts of distributed ledgers and how they can be used to ensure integrity for arbitrary data.
Retrieving passages instead of whole documents can help professionals to acquire new information faster. This is important in domains where time for research is limited and expensive. For example a medical doctor at
an hospital has usually less than an hour per day to look up fresh information for rare cases. We present an approach for retrieving relevant passages in a document collection which leverages two orthogonal semantic embeddings. For this occasion we demonstrate a first prototype implementation as described in . In this talk, we give an overview of our approach to learn a joint vector space representation of these embeddings. We plan to exploit this model to further improve passage and document retrieval tasks.
Eye tracking as a tool to quantify user attention plays a major role in research and application design. For Web page usability, it has become a prominent measure to assess which sections of a Web page are read, glanced or skipped. Such assessments primarily depend on the mapping of gaze data to a Web page representation. However, current representation methods, a virtual screenshot of the Web page or a video recording of the complete interaction session, suffer either from accuracy or scalability issues. We present a method that identifies fixed elements on Web pages and combines user viewport screenshots in relation to fixed elements for an enhanced representation of the page.
In this talk, the DFG funded project Cognitive Reasoning (CoRg) will be introduced. CoRg aims at the construction of a cognitive computing system. Cognitive computing addresses problems characterized by ambiguity and uncertainty, meaning that it is used to handle problems humans are confronted with in everyday life. When developing a cognitive computing system which is supposed to act human-like one cannot rely on utomated theorem proving techniques alone, since humans performing commonsense reasoning do not obey the rules of classical logics. This causes humans to be susceptible to logical fallacies, but on the other hand to draw useful conclusions automated reasoning systems are incapable of.
One of the most important concerns in the field of adaptive information systems is to model human behaviour in order to offer personalisation and recommendation. This usually involves implicit or explicit knowledge about a user's preferences and behavioural patterns.
In this talk, I report on the lack of reliability of explicit user feedback and its interpretation in the light of system evaluation. By using probabilistic perspectives as used in metrology and physics as well as neuroscientific theories of the Bayesian brain, I will introduce novel user models with more empathy for the human nature. By means of user experiments and simulations, I will show that this information can be used to improve the standard collaborative filtering.
Formal Concept Analysis is a mathematically well-founded theory used for (among others) computing concept lattices from data. However, concept lattices induced via FCA tend to be overwhelming in size and complexity, potentially leading to unwarranted overhead in subsequent informational tasks.
This talk discusses a probabilistic approach to deriving concept lattice summarizations that are concise, yet still structurally sound and characteristic of the underlying dataset. The talk concludes in an outlook into future research directions.
While semantic data models are increasingly relevant for scientific and business tasks, working with semantic data still remains complex and error-prone. This, in part, is due to the inadequate integration of related technologies in common programming languages. The research language λDL was developed to remedy this concern, by introducing static checks: It uses description logics, the underlying formalism of OWL ontologies, to provide a type system for semantic data. This thesis is based on λDL and aims to transfer the approach to the functional programming language Scala and the widely used semantic query language SPARQL.
Computational models of argumentation are logic-based formalisms for knowledge representation that allow for the explicit modelling of automatic reasoning in terms of arguments, counterarguments, and their interplay. In this talk I give an overview on the core formalism in this area, abstract argumentation frameworks, and discuss its algorithmic issues.