Computational Linguistics & Phonetics Computational Linguistics & Phonetics Fachrichtung 4.7 Universität des Saarlandes

Overview
People
Annotation manual
Publications
Data

Situation Entities Project

The project

Linguistic expressions form patterns in discourse. Passages of text can be analyzed in terms of the individuals, concepts, times and situations that they introduce to the discourse. In this work, we focus on situation entities, which are expressed at the clause level. In her work Modes of Discourse (2003, a shorter version can be found here), Carlota Smith distinguishes the following situation entity types:
  • STATE: introducing specific properties of specific individuals to the discourse.
    Carl is a tenacious fellow.
  • EVENT: introducing a specific event to the discourse.
    The lobster won the quadrille.
  • GENERALIZING SENTENCE: reporting regularities related to specific individuals.
    Mary often feeds my cats.
  • GENERIC SENTENCES: making statements about kinds.
    The lion has a bushy tail.
  • ABSTRACT ENTITIES: a type of embedded situation, the clausal complements of verbs of knowledge or belief.
    I know that Mary refused the offer.
Although these categories are clearly distinct from one another on theoretical grounds, in practice it can be difficult to cleanly draw boundaries between them. Our main research questions are:
  1. Assessment of the applicability of classification of situation entity types as described by Smith (2003): to what extent can situations be classified easily, which borderline cases occur, and how do humans perform on this task?
  2. Training, development and evaluation of automatic systems classifying situation entities, as well as sub-tasks which have (partially) been studied by the NLP community, but for which no large annotated corpora are available (for example, automatically predicting the fundamental aspectual class of verbs in context (Friedrich and Palmer, ACL 2014) or the genericity of clauses and noun phrases (Friedrich and Pinkal, ACL 2015)).
  3. To provide a foundation for analysis of the theory of Discourse Modes. Smith (2003) proposes the modes of Narrative, Report, Description, Informative, and Argument at the level of passages. The modes differ by their kind of progression and distribution of situation entity types.

People

  • Annemarie Friedrich (researcher, Ph.D. candidate)
  • Alexis Palmer (researcher, post-doc)
  • Manfred Pinkal (advisor, professor)
  • Melissa Peate Sørensen (annotator, intern)
  • Kleio-Isidora Mavridou (annotator, M.Sc. thesis)
  • Liesa Heuschkel (M.Sc. thesis)
  • Christine Bocionek (annotator)
  • Fernando Ardente (annotator)
  • Damyana Gateva (annotator)
  • Ruth Kühn (past, annotator)
  • Ambika Kirkland (past, annotator)
  • Steffen Witt (past, annotator)

Publications

Annemarie Friedrich and Manfred Pinkal: Automatic recognition of habituals: a three-way classification of clausal aspect. September 2015. In Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing (EMNLP). Lisbon, Portugal. (to appear)

Kleio-Isidora Mavridou, Annemarie Friedrich, Melissa Peate Sørensen, Alexis Palmer and Manfred Pinkal: Linking discourse modes and situation entity types in a cross-linguistic corpus study. September 2015. In Proceedings of Linking Models of Lexical, Sentential and Discourse-level Semantics (LSDSem). Lisbon, Portugal. (to appear)

Annemarie Friedrich and Manfred Pinkal: Discourse-sensitive Automatic Identification of Generic Expressions. August 2015. In Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics (ACL). Beijing, China. (slides)

Annemarie Friedrich, Alexis Palmer, Melissa Peate Sørensen and Manfred Pinkal: Annotating genericity: a survey, a scheme, and a corpus. June 2015. In Proceedings of the 9th Linguistic Annotation Workshop (LAW IX). Denver, Colorado, US. (slides)

Annemarie Friedrich and Alexis Palmer
Situation entity annotation.
August 2014. In Proceedings of the 8th Linguistic Annotation Workshop. Dublin, Ireland. (slides)

Alexis Palmer and Annemarie Friedrich
Genre distinctions and discourse modes: Text types differ in their situation type distributions.
July 2014. In Proceedings of the Workshop on Frontiers and Connections between Argumentation Theory and Natural Language Processing. Forlì-Cesena, Italy.

Annemarie Friedrich and Alexis Palmer
Automatic prediction of aspectual class of verbs in context
June 2014. In Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (ACL). Baltimore, USA. (poster)

Alexis Palmer and Caroline Sporleder
Situation entities and genre distinctions in the Penn Discourse TreeBank (abstract)
TLS 2009, Austin, Texas

Alexis Palmer, Elias Ponvert, Jason Baldridge, and Carlota Smith
A Sequence Model for Situation Entity Classification
ACL 2007, Prague

Alexis Palmer, Jonas Kuhn, and Carlota Smith
Utilization of Multiple Language Resources for Robust Grammar-Based Tense and Aspect Classification
LREC 2004, Lisbon

Annotation manual and other resources


Data (for download)

Please cite the respective papers if you are using these resources. Thank you.
  • AspMASC.csv and AspAmbig.csv: data annotated for aspectual class (see ACL 2014 paper)
  • linguistic-indicators-Gigagword-AFE-XIE.csv: database of linguistic indicator values (see ACL 2014 paper) computed from Gigaword (AFE and XIE sections), extracted using the method described in: Siegel, E. V., & McKeown, K. R. (2000). Learning methods to combine linguistic indicators: Improving aspectual classification and revealing linguistic insights. Computational Linguistics, 26(4), 595-628.
  • WikiGenerics corpus v2.0: annotated for genericity on NP- and clause-level (see LAW 2015 and ACL 2015 papers) and clausal aspect (see EMNLP 2015 paper). MASC data annotated with genericity is going to be available here soon!
  • Implementation of features for ACL 2015 paper is going to be available here soon!