Triple Classification Using Regions and Fine-Grained Entity Typing
Authors: Tiansi Dong, Zhigang Wang, Juanzi Li, Christian Bauckhage, Armin B. Cremers77-85
AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The experiments are based on large datasets derived from the benchmark datasets WN11, FB13, and WN18. Our results show that the performance of the new method is related to the length of the type chain and the quality of pre-trained entity-embeddings, and that performances of long chains with welltrained entity-embeddings outperform other methods in the literature. Source codes and datasets are located at https: //github.com/Gnod Is Nait/mushroom. |
| Researcher Affiliation | Collaboration | Tiansi Dong B-IT, University of Bonn, Germany dongt@bit.uni-bonn.de Zhigang Wang Aisino Corporation, China wangzhigang@aisino.com Juanzi Li DCST, Tsinghua University, China lijuanzi@tsinghua.edu.cn Christian Bauckhage,1,2 Armin B. Cremers1 1B-IT, University of Bonn, Germany 2Fraunhofer IAIS, Germany {bauckhag, abc}@bit.uni-bonn.de |
| Pseudocode | Yes | Algorithm 1: construct n-ball embeddings; Algorithm 2: Triple predict(x, h, r, KG, γ, EV) |
| Open Source Code | Yes | Source codes and datasets are located at https: //github.com/Gnod Is Nait/mushroom. |
| Open Datasets | Yes | The experiments are based on large datasets derived from the benchmark datasets WN11, FB13, and WN18. ... WN11-nball datasets and pre-trained entity-embeddings are free for public access3. ... FB13-nball datasets and pre-trained entity-embeddings are free for downloads2. ... WN18-nball datasets and pre-trained entity-embeddings are free for downloads4. |
| Dataset Splits | Yes | WN11 dataset has 11 relations, 38,696 entities, 112,581 training Triples, 2,609 valid Triples, 21,088 testing Triples. ... WN11-nball dataset has 5 relations, totaling 94,472 training Triples (91,888 from WN11 training Triples, 2,584 from WN11 valid Triples), and 20,495 testing Triples; |
| Hardware Specification | No | The paper describes experiments and datasets but does not specify any hardware details like GPU/CPU models or memory used for running the experiments. |
| Software Dependencies | No | The paper mentions software like 'TEKE model' and 'Trans E' but does not provide specific version numbers for any software dependencies. |
| Experiment Setup | Yes | We update the final radius rdhr with a ratio γ to maximize performances of predicting results, as described in Algorithm 2. ... We expand the range of γ from 0.6 to 2.3, to see whether smaller γ can contribute to precision. |