Curvature-Adaptive Embedding of Hyperbolic Geographic 知识图谱
收藏资源简介：
双曲地理知识图谱的曲率自适应嵌入<pre>The source codes that support the paper <b>'</b>Curvature-Adaptive Embedding of Hyperbolic Geographic Knowledge Graph<b>'</b><br><br>This program provides the implementation of our CAH-GKGE as described in our paper.</pre>要求Python 3.7 版PyTorch 1.10.0 & CUDA 11.8运行命令：执行<code>.sh: bash ./CAH-GKGE/model/script/metatrain.sh</code><pre>dataset='10_m' # '1_m', '2_m','3_m','4_m','5_m','6_m','7_m','8_m','9_m','10_m','11_m','12_m','13_m'<br>kge='ComplEx' # 'TransE', 'DistMult', 'ComplEx''<br>CUDA_VISIBLE_DEVICES=6 nohup python main.py --data_name ${dataset} --name ${dataset,,}_${kge,,} \<br> --step meta_train --kge ${kge} > result/train_ComplEx_10_m.out 2>&1 &</pre>
双曲地理知识图谱（Hyperbolic Geographic Knowledge Graph）的曲率自适应嵌入（Curvature-Adaptive Embedding） 本项目提供支撑论文《Curvature-Adaptive Embedding of Hyperbolic Geographic Knowledge Graph》的源代码，实现了论文中提出的CAH-GKGE方法。 本程序的运行环境要求为Python 3.7版本、PyTorch 1.10.0及CUDA 11.8。 运行程序需执行如下Shell脚本： `bash ./CAH-GKGE/model/script/metatrain.sh` 可配置参数如下： - `dataset`：可选取值为`'1_m'`、`'2_m'`、`'3_m'`、`'4_m'`、`'5_m'`、`'6_m'`、`'7_m'`、`'8_m'`、`'9_m'`、`'10_m'`、`'11_m'`、`'12_m'`、`'13_m'` - `kge`：可选取值为`'TransE'`、`'DistMult'`、`'ComplEx'` 完整运行指令示例： CUDA_VISIBLE_DEVICES=6 nohup python main.py --data_name ${dataset} --name ${dataset,,}_${kge,,} --step meta_train --kge ${kge} > result/train_ComplEx_10_m.out 2>&1 &
