In community merging part for cis, best strategy is not applied, since cis expect totally including relationship. Extra cost is introduced in current implementation, a better solution can be that, build index for large community, iterate through small community and judge whether v in small community is found in the large community.
Here, graph algorithms are those locality-based overlapping community detection algorithms.
| content | detail |
|---|---|
| src/algorithm | implemented graph algorithms |
| src/algorithm_demo | simple demo source codes to execute graph algorithm program |
| src/parallel_utils | parallel utilities for accelerating computations of graph algorithms |
| src/util | utilities for graph input and pretty printing |
| content | detail |
|---|---|
| scripts/analyze_algo_quality.py | quality analyzer |
| scripts/metrics/link_belong_modularity.py | link belonging modularity |
| content | detail |
|---|---|
| demo_files/demo_graph.csv | toy graph in edge-list format |
| demo_files/demo_result.txt | computation result of connected-iterative-scan algorithm |
- small_datasets/collaboration_edges_input.csv, from uci repo
- small_datasets/karate_edges_input.csv, from uci repo