GraphSC is a parallel secure computation framework that supports graph-parallel programming abstractions resembling GraphLab. GraphSC is suitable for both multi-core and cluster-based computing architectures. Link for the paper.
git clone https://github.com/kartik1507/GraphSC.git
cd GraphSC/bin/
./compile.sh
./runOne.py
e.g. ./runOne.py pr.PageRank 16 2
The above example will run the PageRank example using 2 garblers and 2 evaluators on the same machine. The configuration for running garblers and evaluators on a cluster can be found in machine_spec/. The input files are stored in in/.
In case of any queries, please contact Kartik Nayak (kartik@cs.umd.edu)
1.支持异构图
2.异构图的输入格式需按照./in/HeteroGraph.in，即0表示顶点，其他数字表示边的类型（与其他示例中1表示顶点不同）
3.目前只改造了PageRank
4.支持两方都有数据
5.支持Alice(即guest)获取结果，结果存储在./work_space/jobid/role/result下（运行文件自动生成）
6.待支持多种特征的提取--experiments pr.PageRank,histogram.Histogram
（本地单机测试时，需先启动guest，因为他会先杀死相关进程）
python ./runOneHetero.py --jobid test --experiments pr.PageRank --role guest --input_length 10 --garblers 4 --num_of_edge_type 2
python ./runOneHetero.py --jobid test --experiments pr.PageRank --role host --input_length 20 --garblers 4 --num_of_edge_type 2
########
######################################################################################################################## ########################################################################################################################
构建图 获取交集的n阶邻域展示
图表示，单方计算 in_degree,out_degree,pagerank,betweenness_centralitys,closeness_centrality
实体属性
{
"attrs": [
[
"高品质服务+跨越式成长，长期价值可期" (实体名称),
"评级"(属性),
"优于大市" (属性值)
],
[
"高品质服务+跨越式成长，长期价值可期",
"发布时间",
"2020/2/19"
],
]
}
实体 { "人物": [ "王勃华", "杨光磊", ], "行业": [ "可穿戴设备", "零售", ], }
关系 { "relationships": [ [ "赋能全球生物药研发的 CRO/CDMO 龙头", "涉及", "药明生物" ], [ "赋能全球生物药研发的 CRO/CDMO 龙头", "涉及", "Arcus Biosicence" ], ] }