An Unsupervised Approach to Recognizing Discourse Relations

By Daniel Marcu, Abdessamad Echihabi. Annual Meeting of the Association for Computational Linguistics, 2002

452 citations recorded by the source corpus.

We present an unsupervised approach to recognizing discourse relations of CONTRAST, EXPLANATION-EVIDENCE, CONDITION and ELABORATION that hold between arbitrary spans of texts. We show that discourse relation classifiers trained on examples that are automatically extracted from massive amounts of text can be used to distinguish between some of these relations with accuracies as high as 93%, even when the relations are not explicitly marked by cue phrases.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/d208d98c010f3c1b5e2bc40180b9e2e69b88c289

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Jianfeng Gao
2. Iryna Gurevych
3. Richard Socher
4. Percy Liang
5. Caiming Xiong
6. Mohit Bansal
7. Luke S. Zettlemoyer
8. Milica Gasic

Citations in the corpus

• Finding Contradictions in Text (2008, methodology): “Presently, there exist methods for learning oppositional terms (Marcu and Echihabi, 2002) and paraphrase learn-”
• DisSent: Learning Sentence Representations from Explicit Discourse Relations (2019, background): “This simplification sidesteps a huge body of work in determining the correct relation for a sentence (Marcu and Echihabi, 2002), in describing the wealth of complex structures a discourse can take (Webber et al.”
• Automatic sense prediction for implicit discourse relations in text (2009, background): “Indeed, word pairs form the basic feature of most previous work on classifying implicit relations (Marcu and Echihabi, 2001; BlairGoldensohn et al., 2007; Sporleder and Lascarides, 2008) or the simpler task of predicting which connective should be used to express a relation (Lapata and Lascarides, 2”
• Recognizing Implicit Discourse Relations in the Penn Discourse Treebank (2009, background): “One of the first works that use statistical meth-ods to detect implicit discourse relations is that of Marcu and Echihabi (2002).”
• Parsing Argumentation Structures in Persuasive Essays (2016, background): “2014; Afantenos et al. 2015). However, similar to argumentation structure parsing the main challenge of discourse analysis is to identify implicit discourse relations (Braud and Denis 2014, p. 1694). Marcu and Echihabi (2002) proposed one of the first approaches for identifying implicit discourse rel”
• Automatic detection of arguments in legal texts (2007, background): “Also, Marcu and Echihabi [19] demonstrate that simple features successfully can identify discourse relations, if sufficient training data are available.”
• Discourse Relation Sense Classification Using Cross-argument Semantic Similarity Based on Word Embeddings (2016, background): “More recently Rutherford and Xue (2015) improved inference of implicit discourse relations via classifying explicit discourse connectives, extending prior research (Marcu and Echihabi, 2002; Sporleder and Lascarides, 2008).”
• Counter-fitting Word Vectors to Linguistic Constraints (2016, background): “The most common use of antonymy information is to provide features for systems that detect contradictions or logical entailment (Marcu and Echihabi, 2002; de Marneffe et al., 2008; Zanzotto et al., 2009).”
• Implicit Discourse Relation Identification for Open-domain Dialogues (2019, methodology): “Inspired by (Braud and Denis, 2014; Marcu and Echihabi, 2002), we use two patterns: (Arg 1) (connective) (Arg 2) and (Arg 1).”
• Multiple Aspect Ranking Using the Good Grief Algorithm (2007, methodology): “The effectiveness of these features for recognition of discourse relations has been previously shown by Marcu and Echihabi (2002).”
• Mining Discourse Markers for Unsupervised Sentence Representation Learning (2019, methodology): “This objective is shared with semi-supervised approaches to discourse relation prediction, where automatically extracted explicit instances feed a model targetting implicit instances (Marcu and Echihabi, 2002; Sporleder and Lascarides, 2008; Pitler and Nenkova, 2009; Rutherford and Xue, 2015).”
• DisSent: Sentence Representation Learning from Explicit Discourse Relations (2017, background): “Given this assumption, researchers have attempted to identify the correct relation for a given pair of sentences (Marcu and Echihabi, 2002), or to determine when and why a discourse relation will be made explicit with a discourse marker (Patterson and Kehler, 2013; Yung et al., 2017).”
• Computing Word-Pair Antonymy (2008, background): “Antonyms often indicate the discourse relation of contrast (Marcu and Echihabi, 2002).”
• Identifying Argumentative Discourse Structures in Persuasive Essays (2014, background): “Marcu and Echihabi (2002) present the first approach focused on identifying implicit discourse relations.”