A Pattern-Based Approach to Recognizing Time Expressions

Authors: Wentao Ding, Guanji Gao, Linfeng Shi, Yuzhong Qu6335-6342

AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results show that PTime achieves a very competitive performance as compared with existing state-of-the-art approaches.
Researcher Affiliation Academia Wentao Ding, Guanji Gao, Linfeng Shi, Yuzhong Qu National Key Laboratory for Novel Software Technology, Nanjing University, Nanjing, China {wtding, gjgao, lfshi}@smail.nju.edu.cn, yzqu@nju.edu.cn
Pseudocode Yes Algorithm 1: Algorithm for Pattern Selection; Algorithm 2: Greedy Select
Open Source Code No The detailed results including lists of selected patterns can be found at http://ws.nju.edu.cn/ptime. There is no explicit statement about releasing the source code for the methodology described in this paper.
Open Datasets Yes We evaluate our approach on the Temp Eval-3 (Uz Zaman et al. 2013), the Wiki Wars (Mazur and Dale 2010) and the Tweets (Zhong, Sun, and Cambria 2017). For the Temp Eval-3, we use the training and test sets splits following previous studies (Bethard 2013) i.e. use the Time Bank (Pustejovsky et al. 2003) corpus as the training set and the platinum-annotated corpus as the test set.
Dataset Splits Yes For development, we perform a 10-fold cross-validation on each training dataset.
Hardware Specification Yes running on a personal workstation with an Intel E3-1226 CPU of 3.30GHz.
Software Dependencies No The paper states the implementation is "written by Java and Scala" but does not provide specific version numbers for these languages or any libraries/frameworks used.
Experiment Setup Yes Parameter Settings We grid searched the value of ρ with a step of 0.01 for maximizing the strict match F1 score on each dataset. The values of ρ are set to 0.87, 0.94, 0.94 for Temp Eval-3, Wiki Wars and Tweets respectively.