Covering various Needs in Temporal Annotation: a Proposal of Extension of ISO TimeML that Preserves Upward Compatibility
15003Authors
nan
Abstract
AbstractThis paper reports a critical analysis of the ISO TimeML standard, in the light of several experiences of temporal annotation that were conducted on spoken French. It shows that the norm suffers from weaknesses that should be corrected to fit a larger variety of needs inNLP and in corpus linguistics. We present our proposition of some improvements of the norm before it will be revised by the ISO Committee in 2017. These modifications concern mainly (1) Enrichments of well identified features of the norm: temporal function of TIMEX time expressions, additional types for TLINK temporal relations; (2) Deeper modifications concerning the units or features annotated: clarification between time and tense for EVENT units, coherence of representation between temporal signals (the SIGNAL unit) and TIMEX modifiers (the MOD feature); (3) A recommendation to perform temporal annotation on top of a syntactic (rather than lexical) layer (temporal annotation on a treebank).
Keywords
Full Content
Covering various Needs in Temporal Annotation: a Proposal of Extension of ISO TimeML that Preserves Upward Compatibility Anaïs Lefeuvre-Halftermeyer1,2, Jean-Yves Antoine2, Alain Couillault4, Emmanuel Schang3, Lotfi Abouda3, Agata Savary2, Denis Maurel2, Iris Eshkol-Taravella3, Delphine Battistelli5 (1) Université Paris-Sorbonne, laboratoire STIH 1, rue Victor Cousin, 75005 Paris, France (2) Université François Rabelais de Tours, laboratoire LI 3 place Jean Jaurès, 41000 Blois, France (3) Université d’Orléans, laboratoire LLL, UMR 7270 10 rue de Tours, 45100 Orléans-la-Source, France (4) Université de La Rochelle, laboratoire L3i Avenue Michel Crépeau, 17000 La Rochelle, France (5) Université Paris-Ouest Nanterre & MoDyCo UMR 7114 200, avenue de la République, 92 001 Nanterre Cedex, France E-mail: anais.lefeuvre@paris-sorbonne.fr, jean-yves.antoine@univ-tours.fr, alain.couillault@univ-lr.fr, emmanuel.schang@univ-orleans.fr, lotfi.abouda@univ-orleans.fr, agata.savary@univ-tours.fr, denis.maurel@univ- tours.fr, iris.eshkol@univ-orleans.fr, delphine.battistelli@u-paris10.fr Abstract This paper reports a critical analysis of the ISO TimeML standard, in the light of several experiences of temporal annotation that were conducted on spoken French. It shows that the norm suffers from weaknesses that should be corrected to fit a larger variety of needs in NLP and in corpus linguistics. We present our proposition of some improvements of the norm before it will be revised by the ISO Committee in 2017. These modifications concern mainly (1) Enrichments of well identified features of the norm: temporal function of TIMEX time expressions, additional types for TLINK temporal relations; (2) Deeper modifications concerning the units or features annotated: clarification between time and tense for EVENT units, coherence of representation between temporal signals (the SIGNAL unit) and TIMEX modifiers (the MOD feature); (3) A recommendation to perform temporal annotation on top of a syntactic (rather than lexical) layer (temporal annotation on a treebank). Keywords: ISO-TimeML, temporal annotation, eventualities, temporal relations, time expressions 1. Introduction Corpus annotation is a time-consuming and a costly activity, thus impeding the development of large language resources. In order to favour the spread and the reuse of such valuable data, a crucial recommendation is to follow established annotation standards. Nonetheless, many areas of Natural Language Processing are still suffering from compliance problems between corpora following different annotation schemes. Temporal annotation has avoided this pitfall. Most time- annotated corpora respect a common norm, TimeML, which was proposed as an international standard ISO DIS 24617-1 (ISO, 2009) now included in the ISO SemAF annotation framework. This standard has been applied to a large variety of languages (English, French, Italian, Portuguese, Korean, Romanian, Chinese...) 1 with a low number of language-specific adjustments which do not affect its coherence. This salutary standardisation effort shows, however, some drawbacks, since TimeML was originally meant for rather specific purposes. The main aim of the norm is to increase the performance of question answering systems, thus, few important problems are addressed by the norm, such as event temporal identification, time stamping of events, reasoning with contextually underspecified temporal expressions and reasoning over the persistence of events (Mani et al. 2005). 1 See (Bittar 2011 ; Caselli 2009 ; Costa & Branco 2008 ; Im et al. 2009 ; These original specific motivations explain that TimeML results from …
Content truncated to 500 words
Labeled Text
AbstractThis paper reports a critical analysis of the ISO TimeML standard, in the light of several experiences of temporal annotation that were conducted on spoken French. It shows that the norm suffers from weaknesses that should be corrected to fit a larger variety of needs inNLP and in corpus linguistics. We present our proposition of some improvements of the norm before it will be revised by the ISO Committee in 2017. These modifications concern mainly (1) Enrichments of well identified features of the norm: temporal function of TIMEX time expressions, additional types for TLINK temporal relations; (2) Deeper modifications concerning the units or features annotated: clarification between time and tense for EVENT units, coherence of representation between temporal signals (the SIGNAL unit) and TIMEX modifiers (the MOD feature); (3) A recommendation to perform temporal annotation on top of a syntactic (rather than lexical) layer (temporal annotation on a treebank).
Publication Details
- Paper ID:
- 15003
- Year:
- 2016
- Publication:
- TIPSTER
- Publication Type:
- conf
- DOI:
- nan