Skip to Main content Skip to Navigation

Which aspects of discourse relations are hard to learn? Primitive decomposition for discourse relation classification

Abstract : Discourse relation classification has proven to be a hard task, with rather low performance on several corpora that notably differ on the relation set they use. We propose to decompose the task into smaller, mostly binary tasks corresponding to various primitive concepts encoded into the discourse relation definitions. More precisely, we translate the discourse relations into a set of values for attributes based on distinctions used in the map-pings between discourse frameworks proposed by Sanders et al. (2018). This arguably allows for a more robust representation of discourse relations, and enables us to address usually ignored aspects of discourse relation prediction, namely multiple labels and underspecified annotations. We study experimentally which of the conceptual primitives are harder to learn from the Penn Discourse Treebank English corpus, and propose a correspondence to predict the original labels, with preliminary empirical comparisons with a direct model.
Document type :
Conference papers
Complete list of metadatas

Cited literature [18 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-02374114
Contributor : Chloé Braud <>
Submitted on : Thursday, November 21, 2019 - 1:37:06 PM
Last modification on : Thursday, March 26, 2020 - 6:02:36 PM

File

128_Paper (1).pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02374114, version 1

Citation

Charlotte Roze, Chloé Braud, Philippe Muller. Which aspects of discourse relations are hard to learn? Primitive decomposition for discourse relation classification. Annual SIGdial Meeting on Discourse and Dialogue, Sep 2019, Stockholm, Sweden. ⟨hal-02374114⟩

Share

Metrics

Record views

42

Files downloads

59