Combining a Knowledge Graph and Graph Algorithms to Find Hidden Skills at NASA
National Aeronautics and Space Administration
www.nasa.gov
KSATT - Knowledge, Skills,
Abilities, Tasks and Technology
David Meza
Sr. Data Scientist
Office of the Chief Human Capital Officer
1
Combining Knowledge Graphs and Algorithms to Find
Hidden Skills
Enterprise Knowledge Graphs
• Enterprise Knowledge Graphs help companies create their
specific web of knowledge representing their very own domain. As
a result, they can seamlessly break down data silos to use
information assets in an agile way.
• Brings together structured and unstructured data
• Speedier search enabling faster decisions
• Can connect knowledge graphs across domains
• Algorithms (NLP)
• Text annotation
• Knowledge graph enrichment – Path Finding, Centrality, Community Detection
• Text rank
• Similarity
• Sentiment analysis
Algorithms
Node Similarity
compares a set of nodes based
on the nodes they are
connected to. Two nodes are
considered similar if they share
many of the same neighbors.
Node Similarity computes pair-
wise similarities based on the
Jaccard metric, also known as
the Jaccard Similarity Score.
18
// Compare IT jobs with Program Analyst
CALL gds.graph.create.cypher(
'IT_Program_Knowledge',
'MATCH (n) WHERE (n:Occupation) OR (n:Knowledge) RETURN id(n) AS id',
'MATCH (opm:OPM_Series)-[r:IN_OPM_SERIES]-(occ:Occupation) WHERE
opm.title IN ["GS-2210", "GS-0080", "GS-0340", "GS-0343"] WITH occ, opm
MATCH (occ)-[f:FOUND_IN]-(s:Knowledge) WHERE f.IM > 0.5 RETURN id(occ)
AS source, id(s) AS target'
)
// Stream Reuslts
CALL gds.nodeSimilarity.stream('IT_Program_Knowledge')
YIELD node1, node2, similarity
WHERE gds.util.asNode(node1).title = 'Management Analysts'
RETURN gds.util.asNode(node1).title AS Occupation1,
gds.util.asNode(node2).title AS Occupation2, similarity
ORDER BY similarity DESCENDING, Occupation1, Occupation2
Career Change
19
Occupation1 Occupation2 similarity
Management Analysts Information Technology Project Managers 72.22%
Management Analysts Document Management Specialists 68.75%
Management Analysts Sustainability Specialists 68.42%
Management Analysts Online Merchants 62.50%
Management Analysts Business Continuity Planners 61.11%
Management Analysts Health Informatics Specialists 60.00%
Management Analysts Database Administrators 52.94%
Management Analysts Information Security Analysts 47.06%
Management Analysts Web Administrators 47.06%
Management Analysts Computer Network Support Specialists 44.44%
Rank Skills by Occupation
OPM Occupation SOC Group Element Value
GS-1550
computer and information research
scientists
15-
1111.00
['Basic_Skills
']
understanding written sentences and paragraphs in work
related documents. 4.5
GS-1550
computer and information research
scientists
15-
1111.00
['Basic_Skills
'] using mathematics to solve problems. 4.38
GS-1550
computer and information research
scientists
15-
1111.00
['Basic_Skills
']
giving full attention to what other people are saying, taking time
to understand the points being made, asking questions as
appropriate, and not interrupting at inappropriate times. 4.12
GS-1550
computer and information research
scientists
15-
1111.00
['Basic_Skills
']
communicating effectively in writing as appropriate for the
needs of the audience. 4.12
GS-1550
computer and information research
scientists
15-
1111.00
['Basic_Skills
'] talking to others to convey information effectively. 4.12
22
MATCH (opm:OPMSeries)-[r:IN_OPM_Series]-(occ:Occupation)
WHERE opm.series IN ["GS-2210", "GS-1550", "GS-854", "GS-855"]
WITH occ, opm
MATCH (occ)-[f:Found_In]-(s:Basic_Skills)
WHERE f.datavalue > 3.75
RETURN opm.series AS OPM, occ.title AS Occupation, occ.onet_soc_code AS SOC, labels(s) AS Group, s.description AS Element, f.datavalue AS
Value
ORDER BY Value Descending
Employees Similar to an Occupation Based on Abilities
23
Similarity Employee
35Bradford
29Constance
29Craig
29David
29Eric
29Jason
29Mehti
29Naimah
29Steven
29Ted
Contact Information
David Meza – david.meza-1@nasa.gov
Twitter - @davidmeza1
Linkedin - https://www.linkedin.com/in/davidmeza1/
Github – davidmeza1
Madison Gipson
Website: https://madisongipson.com/
LinkedIn: https://linkedin.com/in/mgip
Github: https://github.com/mgipson
Email: madison.gipson@nasa.gov or mgipson@stetson.edu