Citation
Tim Rocktaschel, Matko Bosnjak, Sameer Singh, Sabastian Riedel. "Low-Dimensional Embeddings of Logic". 2014 Workshop on Semantic Parsing (SP14), ACL, 26, June, 2014.

Abstract
Many machine reading approaches, from shallow information extraction to deep semantic parsing, map natural language to symbolic representations of meaning. Representations such as first-order logic capture the richness of natural language and support complex reasoning, but often fail in practice due to their reliance on logical background knowledge and the difficulty of scaling up inference. In contrast, low-dimensional embeddings (i.e. distributional representations) are efficient and enable generalization, but it is unclear how reasoning with embeddings could support the full power of symbolic representations such as first-order logic. In this proof-ofconcept paper we address this by learning embeddings that simulate the behavior of first-order logic.

Electronic downloads

• http://www.aclweb.org/anthology/W/W14/W14-2409.pdf

• HTML

  Tim Rocktaschel, Matko Bosnjak, Sameer Singh, Sabastian
  Riedel. <a
  href="http://www.terraswarm.org/pubs/315.html"
  >Low-Dimensional Embeddings of Logic</a>, 2014
  Workshop on Semantic Parsing (SP14), ACL, 26, June, 2014.

• Plain text

  Tim Rocktaschel, Matko Bosnjak, Sameer Singh, Sabastian
  Riedel. "Low-Dimensional Embeddings of Logic".
  2014 Workshop on Semantic Parsing (SP14), ACL, 26, June,
  2014.

• BibTeX

  @inproceedings{RocktaschelBosnjakSinghRiedel14_LowDimensionalEmbeddingsOfLogic,
      author = {Tim Rocktaschel and Matko Bosnjak and Sameer Singh
                and Sabastian Riedel},
      title = {Low-Dimensional Embeddings of Logic},
      booktitle = {2014 Workshop on Semantic Parsing (SP14)},
      organization = {ACL},
      day = {26},
      month = {June},
      year = {2014},
      abstract = {Many machine reading approaches, from shallow
                information extraction to deep semantic parsing,
                map natural language to symbolic representations
                of meaning. Representations such as first-order
                logic capture the richness of natural language and
                support complex reasoning, but often fail in
                practice due to their reliance on logical
                background knowledge and the difficulty of scaling
                up inference. In contrast, low-dimensional
                embeddings (i.e. distributional representations)
                are efficient and enable generalization, but it is
                unclear how reasoning with embeddings could
                support the full power of symbolic representations
                such as first-order logic. In this proof-ofconcept
                paper we address this by learning embeddings that
                simulate the behavior of first-order logic.},
      URL = {http://terraswarm.org/pubs/315.html}
  }
  

Posted by Barb Hoversten on 19 May 2014.
Groups: services