The ACL Test-of-Time awards recognize two papers each year for their lasting
impact on the Computational Linguistics field: one paper from 25 years earlier,
and one paper from 10 year earlier.
2019 is the first year that these awards are officially presented by the ACL,
building on a grass-roots initiative at NAACL 18.
I am delighted to announce that the following two papers have respectively won
the 1994 and 2009 ToT award. The motivation for the honor is also included.
The 2019 winner of the 1994 Test-of-Time Award is:
Bernard Merialdo. Tagging English Text with a Probabilistic Model.
Computational Linguistics 20(2), pp. 155–171
Motivation: [Merialdo 94] is one of the very first papers to compare training of
probabilistic models (for POS tagging) using a mix of labeled / unlabelled data. It
inspired a generation of researchers to investigate whether probabilistic models
could be estimated from unannotated data. Its elegant experimental design, and
clear presentation of a negative result make it pedagogically very useful.
The 2019 winner of the 2009 Test-of-Time Award is:
Theresa Wilson, Janyce Wiebe, Paul Hoffmann. Recognizing Contextual
Polarity: An Exploration of Features for Phrase-Level Sentiment Analysis.
Computational Linguistics 35(3), pp. 399–433
Motivation: [Wilson et al 09] is a seminal paper on sentiment analysis. It
introduced contextual polarity of sentiment terms, including neutral. It provided a
comprehensive analysis of features (word-level, dependency, sentence-level,
topic) that explain and predict contextual meaning. The corpus, nuanced
approach to polarity and inventory of features have greatly affected subsequent
work.
Barbara Di Eugenio
ACL Conference Officer