LarKC Inference Strategy and Architecture Sample Clauses

LarKC Inference Strategy and Architecture. In LarKC, massive, distributed and necessarily incomplete reasoning is performed over web-scale knowledge sources. Massive inference is achieved by distributing problems across heterogeneous computing resources and coordinated by the LarKC platform, shown at right in Figure 1. Some of the distributed computational resources will run highly coupled, high performance inference on local parallel hardware before communicating results back to the distributed computation. With Web-scale inference, complete information is an empty hope and the distributed computation shown at left includes some failed computations that do not thwart the entire problem solving task. The right side of Figure 1 illustrates the architecture that achieves this. LarKC allocates resources strategically and tactically to: 1) Retrieve raw content and assertions that may contribute to a solution, 2) Abstract that information into the forms needed by its heterogeneous reasoning methods, 3) Select the most promising approaches to try first, 4) Reason, using multiple deductive, inductive, abductive, and probabilistic means to move closer to a solution given the selected methods and data, and 5) Decide whether sufficiently many, sufficiently accurate and precise solutions have been found, and, if not, whether it’s worth trying harder. This basic problem framework is supplied as a plug-in architecture, allowing intra-consortium and extra-consortium researchers and users to experiment with improvements to automated reasoning. This architecture will enable the productive yet frictionless interaction between these disciplines: Plugin Component Based on results in... RETRIEVAL Information Retrieval, Cognition, … ABSTRACTION Machine Learning, Ontology,… SELECTION Statistics, Machine Learning, Cognition, … REASONING Logic, Probabilistic inference, … DECIDING Economics, Computing, Decision theory… Why us The consortium has the required expertise in the necessary technological and scientific areas: Semantic Web, ontologies, inference (both logic-based and probabilistic), distributed computing, information retrieval, cognition, machine learning, and databases. All scientific partners have outstanding scientific reputation, and some of them are leaders in their respective fields. The consortium has the required engineering power provided by Ontotext and HLRS. The consortium has the required exploitation expertise with both innovative SMEs and large corporations (ICT and pharmaceutical). The consortium ha...