Active Learning for Black-Box Semantic Role Labeling with Neural Factors

Authors: Chenguang Wang, Laura Chiticariu, Yunyao Li

IJCAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experimental results demonstrate the effectiveness of both this new active learning framework and the neural query strategy model.
Researcher Affiliation Industry Chenguang Wang, Laura Chiticariu, Yunyao Li IBM Research Almaden chenguang.wang@ibm.com, {chiti, yunyaoli}@us.ibm.com
Pseudocode Yes Algorithm 1 ACTIVESRL : Active Learning for Black-box SRL Model. Input: Labeled training data Dl train, labeled test data Dl test, unlabeled data Du, minimum accuracy change threshold minδ, maximum number of iterations max Iter. Output: An SRL model L srl.
Open Source Code No The paper provides links to third-party SRL models (MATE and CLEAR) that it uses, but there is no explicit statement or link indicating that the authors' own code for ACTIVESRL or QUERYM is open-source or publicly available.
Open Datasets Yes Datasets We conduct all the experiments based on CoNLL2009 shared task for English [Hajiˇc et al., 2009].
Dataset Splits Yes We split the training set into two equal portions: denoted as TRAIN and DEV, and denote the in-domain and out-of-domain test sets in the CoNLL-2009 shared task as TESTid and TESTod.
Hardware Specification No The paper does not provide specific details about the hardware used for running the experiments, such as GPU or CPU models.
Software Dependencies No The paper mentions models like Skip-gram, TransE, and ASGD, but it does not provide specific version numbers for any software libraries, programming languages (other than Python generally), or development environments used for implementation.
Experiment Setup Yes n is set as |DEV| / max Iter. We also empirically set minδ as 0.0001 and max Iter as 10 [Settles, 2010] in line 5 of Algorithm 1. The same parameter values apply to all the above methods.