Large-scale complex reasoning with semantics: Approaches and challenges

Abstract

Huge amounts of data are generated by sensor readings, social media and databases. Such data introduce new challenges due to their volume and variety, and thus, new techniques are required for their utilization. We believe that reasoning can facilitate the extraction of new and useful knowledge. In particular, we may apply reasoning in order to make and support decisions, clean noisy data and derive high-level information from low-level input data. In this work we discuss the problem of large-scale reasoning over incomplete or inconsistent information, with an emphasis on nonmonotonic reasoning. We outline previous work, challenges and possible solutions, both over MapReduce and alternative high performance computing infrastructures.