Markov logic networks forsituated incremental natural language understanding

Casey Kennington; David Schlangen

Markov logic networks forsituated incremental natural language understanding

Casey Kennington
, David Schlangen

Computer Science Department

Bielefeld University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

10 Scopus citations

Abstract

We present work on understanding natural languagein a situated domain, that is, languagethat possibly refers to visually present entities,in an incremental, word-by-word fashion. Such type of understanding is required in conversationalsystems that need to act immediatelyon language input, such as multi-modalsystems or dialogue systems for robots. Weexplore a set of models specified as MarkovLogic Networks, and show that a model thathas access to information about the visual contextof an utterance, its discourse context, aswell as the linguistic structure of the utteranceperforms best. We explore its incrementalproperties, and also its use in a joint parsingand understanding module. We concludethat MLNs offer a promising framework forspecifying such models in a general, possiblydomain-independent way.

Original language	English
Title of host publication	SIGDIAL 2012 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Proceedings of the Conference
Pages	314-323
Number of pages	10
ISBN (Electronic)	9781937284442
State	Published - 2012
Event	13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, SIGDIAL 2012 - Seoul, Korea, Republic of Duration: 5 Jul 2012 → 6 Jul 2012

Publication series

Name	SIGDIAL 2012 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Proceedings of the Conference

Conference

Conference	13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, SIGDIAL 2012
Country/Territory	Korea, Republic of
City	Seoul
Period	5/07/12 → 6/07/12

Cite this

Kennington, C., & Schlangen, D. (2012). Markov logic networks forsituated incremental natural language understanding. In SIGDIAL 2012 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Proceedings of the Conference (pp. 314-323). (SIGDIAL 2012 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Proceedings of the Conference).

Kennington, Casey ; Schlangen, David. / Markov logic networks forsituated incremental natural language understanding. SIGDIAL 2012 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Proceedings of the Conference. 2012. pp. 314-323 (SIGDIAL 2012 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Proceedings of the Conference).

@inproceedings{733b1e40ac28439abae0a20c3c42ccc8,

title = "Markov logic networks forsituated incremental natural language understanding",

abstract = "We present work on understanding natural languagein a situated domain, that is, languagethat possibly refers to visually present entities,in an incremental, word-by-word fashion. Such type of understanding is required in conversationalsystems that need to act immediatelyon language input, such as multi-modalsystems or dialogue systems for robots. Weexplore a set of models specified as MarkovLogic Networks, and show that a model thathas access to information about the visual contextof an utterance, its discourse context, aswell as the linguistic structure of the utteranceperforms best. We explore its incrementalproperties, and also its use in a joint parsingand understanding module. We concludethat MLNs offer a promising framework forspecifying such models in a general, possiblydomain-independent way.",

author = "Casey Kennington and David Schlangen",

note = "Publisher Copyright: {\textcopyright}2012 Association for Computational Linguistics.; 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, SIGDIAL 2012 ; Conference date: 05-07-2012 Through 06-07-2012",

year = "2012",

language = "English",

series = "SIGDIAL 2012 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Proceedings of the Conference",

pages = "314--323",

booktitle = "SIGDIAL 2012 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Proceedings of the Conference",

}

Kennington, C & Schlangen, D 2012, Markov logic networks forsituated incremental natural language understanding. in SIGDIAL 2012 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Proceedings of the Conference. SIGDIAL 2012 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Proceedings of the Conference, pp. 314-323, 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, SIGDIAL 2012, Seoul, Korea, Republic of, 5/07/12.

Markov logic networks forsituated incremental natural language understanding. / Kennington, Casey; Schlangen, David.
SIGDIAL 2012 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Proceedings of the Conference. 2012. p. 314-323 (SIGDIAL 2012 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Proceedings of the Conference).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Markov logic networks forsituated incremental natural language understanding

AU - Kennington, Casey

AU - Schlangen, David

PY - 2012

Y1 - 2012

N2 - We present work on understanding natural languagein a situated domain, that is, languagethat possibly refers to visually present entities,in an incremental, word-by-word fashion. Such type of understanding is required in conversationalsystems that need to act immediatelyon language input, such as multi-modalsystems or dialogue systems for robots. Weexplore a set of models specified as MarkovLogic Networks, and show that a model thathas access to information about the visual contextof an utterance, its discourse context, aswell as the linguistic structure of the utteranceperforms best. We explore its incrementalproperties, and also its use in a joint parsingand understanding module. We concludethat MLNs offer a promising framework forspecifying such models in a general, possiblydomain-independent way.

AB - We present work on understanding natural languagein a situated domain, that is, languagethat possibly refers to visually present entities,in an incremental, word-by-word fashion. Such type of understanding is required in conversationalsystems that need to act immediatelyon language input, such as multi-modalsystems or dialogue systems for robots. Weexplore a set of models specified as MarkovLogic Networks, and show that a model thathas access to information about the visual contextof an utterance, its discourse context, aswell as the linguistic structure of the utteranceperforms best. We explore its incrementalproperties, and also its use in a joint parsingand understanding module. We concludethat MLNs offer a promising framework forspecifying such models in a general, possiblydomain-independent way.

UR - https://www.scopus.com/pages/publications/84988014311

M3 - Conference contribution

AN - SCOPUS:84988014311

T3 - SIGDIAL 2012 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Proceedings of the Conference

SP - 314

EP - 323

BT - SIGDIAL 2012 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Proceedings of the Conference

T2 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, SIGDIAL 2012

Y2 - 5 July 2012 through 6 July 2012

ER -

Kennington C, Schlangen D. Markov logic networks forsituated incremental natural language understanding. In SIGDIAL 2012 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Proceedings of the Conference. 2012. p. 314-323. (SIGDIAL 2012 - 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Proceedings of the Conference).

Markov logic networks forsituated incremental natural language understanding

Abstract

Publication series

Conference

Other files and links

Fingerprint

Cite this