Compare commits

...

3 Commits

Author SHA1 Message Date
4321554650 完整配置文件 2024-12-12 16:29:33 +08:00
d2ea109139 生成ss节点和sss节点 2024-12-12 16:29:03 +08:00
babcf04b91 优化 2024-12-12 15:57:25 +08:00
2 changed files with 72 additions and 63 deletions

View File

@ -26,7 +26,7 @@ log:
console: true console: true
sources: sources:
- path: /data/y4/tools/data/user-0s.csv - path: /data/y4/tools/data/users-0.csv
csv: csv:
delimiter: "|" delimiter: "|"
withHeader: true withHeader: true
@ -86,7 +86,7 @@ sources:
type: "STRING" type: "STRING"
index: 16 index: 16
- path: /data/y4/tools/data/post-0s.csv - path: /data/y4/tools/data/posts-0.csv
csv: csv:
delimiter: "|" delimiter: "|"
withHeader: true withHeader: true
@ -140,7 +140,7 @@ sources:
type: "INT" type: "INT"
index: 14 index: 14
- path: /data/y4/tools/data/group-0s.csv - path: /data/y4/tools/data/groups-0.csv
csv: csv:
delimiter: "|" delimiter: "|"
withHeader: true withHeader: true
@ -161,7 +161,7 @@ sources:
type: "STRING" type: "STRING"
index: 3 index: 3
- path: /data/y4/tools/data/organization-0s.csv - path: /data/y4/tools/data/organizations-0.csv
csv: csv:
delimiter: "|" delimiter: "|"
withHeader: true withHeader: true
@ -185,7 +185,7 @@ sources:
type: "STRING" type: "STRING"
index: 4 index: 4
- path: /data/y4/tools/data/task-0s.csv - path: /data/y4/tools/data/tasks-0.csv
csv: csv:
delimiter: "|" delimiter: "|"
withHeader: true withHeader: true
@ -284,25 +284,25 @@ sources:
type: "STRING" type: "STRING"
index: 0 index: 0
# - path: /data/y4/tools/data/post_repost_post-0.csv - path: /data/y4/tools/data/post_repost_post-0.csv
# csv: csv:
# delimiter: "|" delimiter: "|"
# withHeader: true withHeader: true
# edges: edges:
# - name: post_repost_post - name: post_repost_post
# mode: INSERT mode: INSERT
# src: src:
# id: id:
# type: "STRING" type: "STRING"
# index: 2 # src index: 2 # src
# dst: dst:
# id: id:
# type: "STRING" type: "STRING"
# index: 1 # dst index: 1 # dst
# props: props:
# - name: "extra_prop" - name: "extra_prop"
# type: "STRING" type: "STRING"
# index: 0 index: 0
- path: /data/y4/tools/data/post_cite_post-0.csv - path: /data/y4/tools/data/post_cite_post-0.csv
csv: csv:
@ -324,25 +324,25 @@ sources:
type: "STRING" type: "STRING"
index: 0 index: 0
# - path: /data/y4/tools/data/user_post_post-0.csv - path: /data/y4/tools/data/user_post_post-0.csv
# csv: csv:
# delimiter: "|" delimiter: "|"
# withHeader: true withHeader: true
# edges: edges:
# - name: user_post_post - name: user_post_post
# mode: INSERT mode: INSERT
# src: src:
# id: id:
# type: "STRING" type: "STRING"
# index: 2 # src index: 2 # src
# dst: dst:
# id: id:
# type: "STRING" type: "STRING"
# index: 1 # dst index: 1 # dst
# props: props:
# - name: "extra_prop" - name: "extra_prop"
# type: "STRING" type: "STRING"
# index: 0 index: 0
- path: /data/y4/tools/data/user_comment_user-0.csv - path: /data/y4/tools/data/user_comment_user-0.csv
csv: csv:
@ -544,22 +544,22 @@ sources:
type: "STRING" type: "STRING"
index: 0 index: 0
# - path: /data/y4/tools/data/user_join_organization-0.csv - path: /data/y4/tools/data/user_join_organization-0.csv
# csv: csv:
# delimiter: "|" delimiter: "|"
# withHeader: true withHeader: true
# edges: edges:
# - name: user_join_organization - name: user_join_organization
# mode: INSERT mode: INSERT
# src: src:
# id: id:
# type: "STRING" type: "STRING"
# index: 2 # src index: 2 # src
# dst: dst:
# id: id:
# type: "STRING" type: "STRING"
# index: 1 # dst index: 1 # dst
# props: props:
# - name: "extra_prop" - name: "extra_prop"
# type: "STRING" type: "STRING"
# index: 0 index: 0

View File

@ -341,15 +341,23 @@ public class GeneratorTestData {
int i = 0; int i = 0;
List<String> edgeLists = Arrays.asList("post_in_task", "user_participate_task", "user_join_group", "user_join_organization"); List<String> edgeLists = Arrays.asList("post_in_task", "user_participate_task", "user_join_group", "user_join_organization");
for (String src : srcList) { for (String src : srcList) {
i++;
if (i % 1000 == 0) { if (i % 1000 == 0) {
log.info("edge_type: {}-{}, {}",edgeType, count, i); log.info("edge_type: {}-{}, {}",edgeType, count, i);
} }
List<String> dsts; List<String> dsts;
if (i == 0 && !edgeLists.contains(edgeType)) { if (i == 0 && !edgeLists.contains(edgeType)) {
// 当索引等于0 选择 1000 个目标节点
dsts = getRandomSample(dstList, 1000);
log.info("super vertex id is {}; dst size is {}", src, dsts.size());
} else if (i == 1 && !edgeLists.contains(edgeType)) {
// 当索引等于0 选择 10001 个目标节点
dsts = getRandomSample(dstList, 1500);
log.info("ssuper vertex id is {}; dst size is {}", src, dsts.size());
} else if (i == 2 && !edgeLists.contains(edgeType)) {
// 当索引等于0 选择 10001 个目标节点 // 当索引等于0 选择 10001 个目标节点
dsts = getRandomSample(dstList, 10001); dsts = getRandomSample(dstList, 10001);
log.info("sssuper vertex id is {}; dst size is {}", src, dsts.size());
} else { } else {
// 否则选择最多 num 个目标节点 // 否则选择最多 num 个目标节点
dsts = getRandomSample(dstList, Math.min(dstList.size(), num)); dsts = getRandomSample(dstList, Math.min(dstList.size(), num));
@ -364,6 +372,7 @@ public class GeneratorTestData {
edge.put("edge_type", edgeType); edge.put("edge_type", edgeType);
edges.add(edge); edges.add(edge);
} }
i++;
} }
return edges; return edges;
} }