In this technical report, we consider the problem of query answering
over large ontologies.  Traditional reasoning systems may have
problems to deal with large amounts of expressive ontological data
(terminological as well as assertional data) that usually must be kept
in main memory. We propose to overcome this problem with a new
so-called filter and refine paradigm for ontology-based query
answering.  In the filter step, the terminological part of an ontology
is approximated to a less expressive ontology language, e.g., the
description logic DL-Lite, which allows for efficient and complete, but
possibly unsound query answering. A query is first evaluated
w.r.t. this DL-Lite ontology; thus, the set of ontology individuals
is filtered with the help of a DL-Lite query. In the second step, this
set of retrieved individuals is refined further, without having to
perform reasoning over the whole ontology. For this, a partition-based
approach is investigated; the aim is to compute partitions which are
small enough for main memory reasoning systems. However, the first
step can be performed on secondary memory, exploiting database
technology.  The contribution of this report is twofold: (1) For both
steps, novel algorithms are presented. (2) We evaluate our approach on
real-world multimedia ontologies