Jointly Extracting Multiple Triplets with Multilayer Translation Constraints
Authors: Zhen Tan, Xiang Zhao, Wei Wang, Weidong Xiao7080-7087
AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In experiment, TME exhibits superior performance and achieves an improvement of 37.6% on F1 score over state-of-the-art competitors.5 Experiments and Analysis |
| Researcher Affiliation | Academia | 1Key Laboratory of Science and Technology on Information System Engineering, National University of Defense Technology, China 2School of Computer Science and Engineering, UNSW, Australia 3Collaborative Innovation Center of Geospatial Technology, China 4College of Computer Science and Technology, DGUT, China |
| Pseudocode | No | The paper does not contain structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide concrete access to source code for the methodology described. It only mentions that *other* prior work had released source code. |
| Open Datasets | Yes | Experiments were carried out on two publicly available datasets NYT-single (Riedel, Yao, and Mc Callum 2010) and NYT-multi (statistics in Table 1). |
| Dataset Splits | Yes | We randomly chose 10% sentences in test set as validation set, and the rest was regarded as evaluation set. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running its experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details (e.g., library or solver names with version numbers) needed to replicate the experiment. |
| Experiment Setup | Yes | For parameter setting, we selected the dimension of word embeddings dw among {20, 50, 100, 200}, the dimension of character embeddings dch among {5, 10, 15, 25}, the dimension of capitalization embeddings dc among {1, 2, 5, 10}, the margin γ between positive and negative triplets among {1, 2, 5, 10}, and the weighting hyperparameter λ among {0.2, 0.5, 1, 2, 5, 10, 20, 50}. The dropout ratio was set to 0 or 0.5. Stochastic gradient descent (Amari 1993) was called to optimize the loss function. ... The optimal configurations were λ = 10.0, γ = 2.0, dw = 100, dch = 25, dc = 5, and dropout = 0.5. |