An Implementation Of The Annis System On Top Of A Column-Oriented Main-Memory Database

Annis is a web-based search engine and visualization tool for multi-modal, multi-layer linguistic text corpora [ZRLC09]. Annotations are taken from various tools and merged into a unified corpus with support for conflicting annotations over the same primary data. It is implemented on top of the relational database managements system (RDBMS) PostgreSQL, using a pre/post-order scheme to store the nodes of an ordered directed acyclic graph (ODAG) [Ros11]. This graph encoding is inspired by the XPath Accelerator, originally developed to store XML documents in a RDBMS [GKT04]. However, the Annis data model differs notably from the XML document model: