Skip to content

Commit 7b87037

Browse files
JohTJohT
committed
Introduce anomaly detection pipeline with tuned clustering
1 parent c1d468f commit 7b87037

File tree

5 files changed

+774
-341
lines changed

5 files changed

+774
-341
lines changed

‎domains/anomaly-detection/AnomalyDetectionFeatures.cypher‎

Lines changed: 24 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,24 @@
1+
// Query code unit nodes with their anomaly detection
2+
3+
MATCH (codeUnit)
4+
WHERE $projection_node_label IN labels(codeUnit)
5+
AND codeUnit[$community_property] IS NOT NULL
6+
AND codeUnit.clusteringHDBSCANLabel IS NOT NULL
7+
AND codeUnit.clusteringHDBSCANProbability IS NOT NULL
8+
AND codeUnit.embeddingFastRandomProjectionVisualizationX IS NOT NULL
9+
AND codeUnit.embeddingFastRandomProjectionVisualizationY IS NOT NULL
10+
OPTIONAL MATCH (artifact:Java:Artifact)-[:CONTAINS]->(codeUnit)
11+
WITH *, artifact.name AS artifactName
12+
OPTIONAL MATCH (projectRoot:Directory)<-[:HAS_ROOT]-(proj:TS:Project)-[:CONTAINS]->(codeUnit)
13+
WITH *, last(split(projectRoot.absoluteFileName, '/')) AS projectName
14+
RETURN DISTINCT
15+
coalesce(codeUnit.fqn, codeUnit.globalFqn, codeUnit.fileName, codeUnit.signature, codeUnit.name) AS codeUnitName
16+
,codeUnit.name AS shortCodeUnitName
17+
,elementId(codeUnit) AS nodeElementId
18+
,coalesce(artifactName, projectName) AS projectName
19+
,codeUnit[$community_property] AS communityId
20+
,codeUnit.clusteringHDBSCANLabel AS clusteringLabel
21+
,codeUnit.clusteringHDBSCANProbability AS clusteringProbability
22+
,codeUnit.embeddingFastRandomProjectionVisualizationX AS visualizationX
23+
,codeUnit.embeddingFastRandomProjectionVisualizationY AS visualizationY
24+
,coalesce(codeUnit.centralityPageRank, 0.00001) AS centrality

‎domains/anomaly-detection/Set_Parameters_Manual.cypher‎

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,8 @@
1+
// Example on how to set the parameters for anomaly detection
2+
3+
:params {
4+
"projection_name": "package-anomaly-detection",
5+
"projection_node_label": "Package",
6+
"projection_weight_property": "weight25PercentInterfaces",
7+
"community_property": "communityLeidenIdTuned",
8+
}

‎domains/anomaly-detection/anomalyDetectionPipeline.sh‎

Lines changed: 128 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,128 @@
1+
#!/usr/bin/env bash
2+
3+
# Pipeline that coordinates anomaly detection using the Graph Data Science Library of Neo4j.
4+
# It requires an already running Neo4j graph database with already scanned and analyzed artifacts.
5+
# The results will be written into the sub directory reports/anomaly-detection.
6+
7+
# Note that "scripts/prepareAnalysis.sh" is required to run prior to this script.
8+
9+
# Requires executeQueryFunctions.sh, projectionFunctions.sh, cleanupAfterReportGeneration.sh
10+
11+
# Fail on any error ("-e" = exit on first error, "-o pipefail" exist on errors within piped commands)
12+
set -o errexit -o pipefail
13+
14+
# Overrideable Constants (defaults also defined in sub scripts)
15+
REPORTS_DIRECTORY=${REPORTS_DIRECTORY:-"reports"}
16+
17+
## Get this "scripts/reports" directory if not already set
18+
# Even if $BASH_SOURCE is made for Bourne-like shells it is also supported by others and therefore here the preferred solution.
19+
# CDPATH reduces the scope of the cd command to potentially prevent unintended directory changes.
20+
# This way non-standard tools like readlink aren't needed.
21+
ANOMALY_DETECTION_SCRIPT_DIR=${ANOMALY_DETECTION_SCRIPT_DIR:-$(CDPATH=. cd -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd -P)}
22+
echo "anomalyDetectionPipeline: ANOMALY_DETECTION_SCRIPT_DIR=${ANOMALY_DETECTION_SCRIPT_DIR}"
23+
# Get the "scripts" directory by taking the path of this script and going one directory up.
24+
SCRIPTS_DIR=${SCRIPTS_DIR:-"${ANOMALY_DETECTION_SCRIPT_DIR}/../../scripts"} # Repository directory containing the shell scripts
25+
# Get the "cypher" directory by taking the path of this script and going two directory up and then to "cypher".
26+
CYPHER_DIR=${CYPHER_DIR:-"${ANOMALY_DETECTION_SCRIPT_DIR}/../../cypher"}
27+
28+
# Function to display script usage
29+
usage() {
30+
echo -e "${COLOR_ERROR}" >&2
31+
echo "Usage: $0 [--verbose]" >&2
32+
echo -e "${COLOR_DEFAULT}" >&2
33+
exit 1
34+
}
35+
36+
# Default values
37+
verboseMode="" # either "" or "--verbose"
38+
39+
# Parse command line arguments
40+
while [[ $# -gt 0 ]]; do
41+
key="$1"
42+
value="${2}"
43+
44+
case ${key} in
45+
--verbose)
46+
verboseMode="--verbose"
47+
;;
48+
*)
49+
echo -e "${COLOR_ERROR}anomalyDetectionPipeline: Error: Unknown option: ${key}${COLOR_DEFAULT}" >&2
50+
usage
51+
;;
52+
esac
53+
shift || true # ignore error when there are no more arguments
54+
done
55+
56+
# Define functions to execute a cypher query from within a given file (first and only argument) like "execute_cypher"
57+
source "${SCRIPTS_DIR}/executeQueryFunctions.sh"
58+
59+
# Define functions to create and delete Graph Projections like "createUndirectedDependencyProjection"
60+
source "${SCRIPTS_DIR}/projectionFunctions.sh"
61+
62+
# Create report directory
63+
REPORT_NAME="anomaly-detection"
64+
FULL_REPORT_DIRECTORY="${REPORTS_DIRECTORY}/${REPORT_NAME}"
65+
mkdir -p "${FULL_REPORT_DIRECTORY}"
66+
67+
# Query Parameter key pairs for projection and algorithm side
68+
PROJECTION_NAME="dependencies_projection"
69+
ALGORITHM_PROJECTION="projection_name"
70+
71+
PROJECTION_NODE="dependencies_projection_node"
72+
ALGORITHM_NODE="projection_node_label"
73+
74+
PROJECTION_WEIGHT="dependencies_projection_weight_property"
75+
ALGORITHM_WEIGHT="projection_weight_property"
76+
77+
# Code independent algorithm parameters
78+
COMMUNITY_PROPERTY="community_property=communityLeidenIdTuned"
79+
80+
# Run the anomaly detection pipeline.
81+
#
82+
# Required Parameters:
83+
# - projection_name=...
84+
# Name prefix for the in-memory projection name. Example: "package-anomaly-detection"
85+
# - projection_node_label=...
86+
# Label of the nodes that will be used for the projection. Example: "Package"
87+
# - projection_weight_property=...
88+
# Name of the node property that contains the dependency weight. Example: "weight"
89+
anomaly_detection_pipeline() {
90+
91+
time "${ANOMALY_DETECTION_SCRIPT_DIR}/tunedLeidenCommunityDetection.py" "${@}" ${verboseMode}
92+
time "${ANOMALY_DETECTION_SCRIPT_DIR}/tunedNodeEmbeddingClustering.py" "${@}" ${verboseMode}
93+
94+
local nodeLabel
95+
nodeLabel=$( extractQueryParameter "projection_node_label" "${@}" )
96+
execute_cypher "${ANOMALY_DETECTION_SCRIPT_DIR}/AnomalyDetectionFeatures.cypher" "${@}" > "${FULL_REPORT_DIRECTORY}/${nodeLabel}AnomalyDetection.csv"
97+
}
98+
99+
# -- Java Artifact Node Embeddings -------------------------------
100+
101+
if createUndirectedDependencyProjection "${PROJECTION_NAME}=artifact-anomaly-detection" "${PROJECTION_NODE}=Artifact" "${PROJECTION_WEIGHT}=weight"; then
102+
anomaly_detection_pipeline "${ALGORITHM_PROJECTION}=artifact-anomaly-detection" "${ALGORITHM_NODE}=Artifact" "${ALGORITHM_WEIGHT}=weight" "${COMMUNITY_PROPERTY}"
103+
fi
104+
105+
# -- Java Package Node Embeddings --------------------------------
106+
107+
if createUndirectedDependencyProjection "${PROJECTION_NAME}=package-anomaly-detection" "${PROJECTION_NODE}=Package" "${PROJECTION_WEIGHT}=weight25PercentInterfaces"; then
108+
anomaly_detection_pipeline "${ALGORITHM_PROJECTION}=package-anomaly-detection" "${ALGORITHM_NODE}=Package" "${ALGORITHM_WEIGHT}=weight25PercentInterfaces" "${COMMUNITY_PROPERTY}"
109+
fi
110+
111+
# -- Java Type Node Embeddings -----------------------------------
112+
113+
if createUndirectedJavaTypeDependencyProjection "${PROJECTION_NAME}=type-anomaly-detection"; then
114+
anomaly_detection_pipeline "${ALGORITHM_PROJECTION}=type-anomaly-detection" "${ALGORITHM_NODE}=Type" "${ALGORITHM_WEIGHT}=weight" "${COMMUNITY_PROPERTY}"
115+
fi
116+
117+
# -- Typescript Module Node Embeddings ---------------------------
118+
119+
if createUndirectedDependencyProjection "${PROJECTION_NAME}=typescript-module-embedding" "${PROJECTION_NODE}=Module" "${PROJECTION_WEIGHT}=lowCouplingElement25PercentWeight"; then
120+
anomaly_detection_pipeline "${ALGORITHM_PROJECTION}=typescript-module-embedding" "${ALGORITHM_NODE}=Module" "${ALGORITHM_WEIGHT}=lowCouplingElement25PercentWeight" "${COMMUNITY_PROPERTY}"
121+
fi
122+
123+
# ---------------------------------------------------------------
124+
125+
# Clean-up after report generation. Empty reports will be deleted.
126+
source "${SCRIPTS_DIR}/cleanupAfterReportGeneration.sh" "${FULL_REPORT_DIRECTORY}"
127+
128+
echo "anomalyDetectionPipeline: $(date +'%Y-%m-%dT%H:%M:%S%z') Successfully finished."

0 commit comments

Comments
 (0)