ISSN 2071-8594

Russian academy of sciences

Editor-in-Chief

Gennady Osipov

E.А. Suleymanova, I.V. Trofimov “A Year Ago”: on Normalisation of one Type of Temporal Expressions

Abstract.

The paper addresses a particular issue within the framework of temporal information normalisation. It focuses on time expressions of the kind ‘three years ago’, ‘two months later’, ‘a week ago’ etc. According to TimeML, the international standard for time markup and annotation, such expressions are to be normalised as dates. We conducted an experiment in order to find out how people interpret time expressions of this kind, and to what extent normalising such expressions as dates is consistent with human judgments. We used two questionnaires to collect human judgments, and then used the answers to create ‘model values’ in the form of membership functions. We implemented four modes of normalisation to reproduce the state-of-the-art method. Evaluation of these normalisation modes on our test data showed, in most cases, great discrepancies between the normalisation results and the ‘model values’. This might be an indication that alternative methods of normalisation would be appropriate. Experiment analysis revealed some normalisation patterns that could serve as a basis for an alternative solution.

Keywords:

natural language processing, temporal information extraction, temporal expressions, timex normalisation, absolute value of relative temporal expressions, deictic and anaphoric temporal references,
TimeML.

PP. 27-39.

DOI 10.14357/20718594200103

References

1. Suleymanova E. A. 2016. O dvukh vidakh tekstovykh vremennykh koordinat [On two types of time-referring expressions]. Programmnye sistemy: teoriya i prilozheniya [Program systems: theory and applications]. 4(31): 209–229. Available at: http://psta.psiras.ru/read/psta2016_4_209-229.pdf
2. L. Ferro, L. Gerber, I. Mani, B. Sundheim, and G. Wilson. 2005. TIDES 2005 standard for the annotation of temporal expressions. Technical report, MITRE, September. https://www.ldc.upenn.edu/sites/www.ldc.upenn.edu/files/english-timex2-guidelines-v0.1.pdf
3. Roser Saur´ı, Jessica Littman, Bob Knippen, Robert Gaizauskas, Andrea Setzer, and James Pustejovsky. TimeML Annotation Guidelines. Version 1.2.1. January 31, 2006 https://catalog.ldc.upenn.edu/docs/LDC2006T08/timeml_annguide_1.2.1.pdf
4. Guidelines for Temporal Expression Annotation for English for TempEval 2010. TimeML Working Group August 14, 2009 http://www.timeml.org/tempeval2/tempeval2-trial/guidelines/timex3guidelines-072009.pdf
5. Tissot, H., Del Fabro, M.D., Derczynski, L. et al. Knowl Inf Syst (2019) 61: 1361. https://doi.org/10.1007/s10115-019-01338-1.