Graphs of Semantically Associated Relations

A sar-graph is a graph containing linguistic knowledge at syntactic and lexical semantic levels for a given language and target relation. The nodes in a sar-graph are either semantic arguments of a target relation or content words (to be more exact, their word senses) needed to express/recognize an instance of the target relation. The nodes are connected by syntactic dependency structure relations and, implicitly via BabelNet, lexical semantic relations. A definition can be found in (Uszkoreit and Xu, 2013).

The current sar-graph release (version 3.0) provides the results of a long-term pattern verification effort (Hennig et al., 2015). Furthermore, the results of resource linking endeavors are included (Krause et al., 2015), i.e., we have added links to WordNet (for sar-graph vertices) and to FrameNet (for patterns).

Sar-graphs are useful for relation extraction, question answering, textual entailment, and summarization, as well as for related downstream applications like computer-assisted language learning (Ai et al., 2015).

The development of sar-graphs is partially supported by

• the German Federal Ministry of Education and Research (BMBF) through the projects Deependance (contract 01IW11003) and All Sides (contract 01IW14002)
  
• the project LUcKY, a Google Focused Research Award for Natural Language Understanding:
  Hans Uszkoreit (DFKI and Saarland University), Feiyu Xu (DFKI and Saarland University) and Roberto Navigli (Sapienza University of Rome) for “Language Understanding cum Knowledge Yield”

Sar-graphs are a result of a collaboration between the DFKI LT-Lab and the BabelNet group at Sapienza University of Rome.