Building Dynamic Knowledge Graphs from Text using Machine Reading Comprehension
Authors: Rajarshi Das, Tsendsuren Munkhdalai, Xingdi Yuan, Adam Trischler, Andrew McCallum
ICLR 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | On two comprehension tasks from the recently proposed PROPARA dataset (Dalvi et al., 2018), our model achieves state-of-the-art results. The model also outperforms previous approaches on the RECIPES dataset (Kiddon et al., 2015), which suggests it may apply broadly to procedural text. We evaluate our model (KG-MRC) on the above two PROPARA tasks and find that the same model significantly outperforms the previous state of the art. |
| Researcher Affiliation | Collaboration | Rajarshi Das 1, Tsendsuren Munkhdalai2, Xingdi Yuan2, Adam Trischler2, Andrew Mc Callum1 1College of Information and Computer Sciences University of Massachusetts, Amherst {rajarshi, mccallum}@cs.umass.edu 2Microsoft Research Montr eal Montr eal, Qu ebec, Canada {tsendsuren.munkhdalai,eric.yuan, adam.trischler}@microsoft.com |
| Pseudocode | No | The paper contains mathematical equations but does not include any explicit pseudocode or algorithm blocks. |
| Open Source Code | No | The paper refers to an "official evaluation pipeline" on GitHub but does not provide a link to the source code for their own model, KG-MRC, nor does it state that their code is being released. |
| Open Datasets | Yes | We evaluate KG-MRC on the recently released PROPARA dataset (Dalvi et al., 2018)... Bosselut et al. (2018) recently released the RECIPES dataset... |
| Dataset Splits | Yes | Table 1: Statistics of PROPARA. # train/#dev/#test 391/43/54 |
| Hardware Specification | No | The paper mentions implementation details, including the use of PyTorch, but does not specify any hardware details such as GPU models, CPU types, or memory. |
| Software Dependencies | Yes | word embeddings are initialized with Fast Text embeddings (Joulin et al., 2016); ... The model is implemented using Py Torch (Paszke et al., 2017). |
| Experiment Setup | Yes | In all experiments, the word embeddings are initialized with Fast Text embeddings (Joulin et al., 2016); we use a document LSTM with two layers, the number of hidden units in each layer is 64. We apply dropout rate of 0.4 in all recurrent layers, and 0.3 in all other layers. The number of recurrent graph layers were set to (L = 2). The hidden unit size for the recurrent graph component was set to 64. During training, the mini-batch size is 8. We use adam (Kingma & Ba, 2014) as the step rule for optimization, The learning rate is set to 0.002. |