A Knowledge Graph Approach to Satisfying Regional Workforce Education Needs

Posted August 10, 2022 | Technology | Amplify
In this issue:

   AMPLIFY  VOL. 35, NO. 7

George Hurlburt details how a KG was used to assist a regional center of a major university system in its course-selection process. The KG helped leaders more clearly see the array of educational pathways from K-12 to community college (CC) coursework that are the results of articulation agreements between universities and CCs. Hurlburt shares five figures from the KG that demonstrate its meaningful visualizations. He also explains how the KG was built, including limiting the number of arcs and emphasizing node unambiguity. Finally, Hurlburt concludes with five key academic relationships and trends that are clearly demonstrated by the regional center’s KG.


Regional centers (RCs) of major university systems typically lack the necessary accreditation to create courses to satisfy local workforce trends. Instead, they must rely on courses developed elsewhere under proper academic oversight, and it’s difficult to attract such courses for myriad reasons. To help solve this dilemma, a knowledge graph (KG) was created to assist a new RC in the course-selection process.

Rather than concentrating on employers and their known unmet needs, this KG examined already-established educational pathways in the region. The results highlight numerous educational drivers that clearly relate to regional workforce needs while showing the subtle differences among the strategies employed and revealing workforce needs among differing counties in the region.

The KG vividly shows numerous richly attributed educational pathways from K-12 to community college (CC) coursework. Each well-defined pathway, complete with collegiate credit offerings, certifications, and valuable career linkages, demonstrates numerous established articulation agreements between universities and CCs.

These agreements include a broad range of upper-level undergraduate offerings with real potential that correspond well with the known workforce needs in the region. The KG is now entering a new phase with further emphasis on extracurricular programs, internships, and apprenticeships to further reinforce the region’s main education drivers.

RCS Are Valuable Workforce Intermediaries

An RC’s nature is largely defined by the community in which it’s placed. Most often, a university system’s RC supports a regional CC focused on responding to local economic development needs. These needs often generate requirements for advanced technical education and frequently lead to full-degree programs. Moreover, the quantity and quality of students feed both workforce expansion and program development. This, in concert with strong industrial support, position RCs as valuable workforce intermediaries.1

Higher education RCs don’t have the authority to act as a college or university, instead relying on fully accredited courses from their respective state university systems or other universities. RCs typically work well with established CCs. This means the RC undergraduate focus typically involves multiple upper-division course partnerships. The development and cultivation of these partnerships is the predominant challenge, not the lack of accreditation.

This article reports on a KG created for a new RC that is uniquely placed in a thriving US technological corridor dedicated to excellence. The notion of autonomous systems, including unmanned vehicles, has a keen economic development interest in this area. This is reinforced by a new RC building dedicated to community cohesion, education, and state-of-the-art research in autonomous vehicles, setting the RC apart from its more traditional counterparts. This emphasis on research is important, as it focuses on science, technology, engineering, and mathematics (STEM), an important aspect of the region’s existing educational framework.

Proximity to a robust CC spanning the three counties that define the region rounds out the equation for a successful RC. The diversity of offerings at the regional CC ensures that regional educational needs can be met, extending well beyond those of the STEM community.

The new RC faces the dilemma of defining manageable educational pathways that: (1) support regional needs with a qualified locally grown and diversified workforce, and (2) provide real opportunity for regional industrial expansion.

An initial survey proved useful from an academic’s professional standpoint but was unwieldy for use in a more broadly based population. Hiring statistics and future job projections served to satisfy immediate needs but lacked sufficient reliability in the longer empirical view when viewed by themselves. This is largely due to a volatile economy further beset by pandemic-induced stress. Other indicators needed to be brought to bear to reinforce the more generalized projections with tangible regional roots. For that, the new RC turned to harvesting reliable regional educational data as reinforced via a KG.

Creating KG Pathway Representation

KGs are valuable instruments when studying complex adaptive systems. Unlike relational database management systems, which require elaborate inflexible schemas, KGs rely on fluid relationships that can come and go over time without major analytic sustainability overhauls.2 

A KG relies on the notion of triples in a subject-predicate-object relationship structure (e.g., “regional center offers electrical engineering”). Spanning these triples allows the development of logic chains that provide traceable pathways, often indirectly linking cause to effect across many nodes and frequently offering multiple alternative paths (predicates) between nodes (subjects and objects).

The RC KG shown in Figures 1-5 drew on readily available public data from the three regional high school (HS) systems, the tri-county CC, the RC itself, and other related national higher education resources. It was built in Neo4j graph database version 1.4.15 using the Cypher query language in the 4.4.6 browser. By design, it permits the construction and visualization of useful graph pathways for existing educational programs of study within the tri-county region.

KG research is emerging from relative infancy. It extends to graph algorithms of all kinds as well as applied graph theoretical mathematics. Advanced KG research deals with the KG for knowledge-aware applications, knowledge acquisition and temporal KGs, and knowledge representation and learning (KRL). The current focus on KRL involves building KG frameworks conducive to applied artificial intelligence and, more specifically, machine learning.3

KGs provide vivid, meaningful visualizations. Although many visualization formalisms and frameworks exist, the age-old art versus science argument is likely applicable.4 Central to both art and science, however, is context, which adds the spice of domain-driven diversification and semantic relevance to the argumentative mix.   

Context is often cited as an essential element to establish KG credibility. Some researchers go so far as to perform extensive Web crawls to develop added triples that lend temporal, spatial, and other contextual value to established triples of interest.5 Others rely on direct representation, reification, higher arity6 representation, and annotations. Ontological-based schemas, of course, provide more precise semantic relationships.7

The RC KG was built using direct representation. Several intentional design steps were taken to ensure context was appropriately addressed. Arcs, the predicates, were limited in number and carefully controlled to ensure relevance. Nodes, the subjects and object entities, were also intentionally kept unambiguous. Where appropriate, triples were added to establish spatial location to the county level within the region. Aggregations were then managed by query.

Each program of study was attributed with several specific properties. Among these properties, enrollment numbers provided valuable quantitative information as to the true viability of potential pathways. When available, feeder courses were also listed.

Educational pathways were defined by progressive programs of study. Program-of-study properties were reinforced by additional related triples showing who is engaged in what and the outcomes of these engagements. The resulting RC KG is, it is hoped, accurate and reflective of existing cultural values of the region served. It is intended to offer a robust backdrop from which future upper-level coursework may be defined based on established regional educational patterns, adding relevance to otherwise speculative future job predictions. The resulting schema appears in Figure 1.

Figure 1. RC KG schema resulting from direct representation
Figure 1. RC KG schema resulting from direct representation 

The ideal pathway extends from public schools to a specific higher-education degree. For example, the pathway for electrical engineering (EE) has its roots in the public schools. Some regional schools adopt nationally accepted programs of study for their curricula. The RC KG defines specific regional HS programs of study, many of which lead to college credit. This credit includes the regional CC, which offers an engineering program leading to an associate of science (AS) degree. The combination of college credits at the HS level and accreditation programs between the CC and four-year higher educational institutions were taken into consideration in constructing what was considered a viable upper-division pathway.

The CC pre-EE offerings are accompanied by internships at the leading technology employer in the region. Working through the school of engineering at the state level, with whom the CC has an articulation agreement, the RC offers the necessary upper-level coursework leading to a bachelor’s degree in EE. In addition to ongoing internships, EE graduates are guaranteed entry- level positions with that employer. The pathway graph in Figure 2 captures all these STEM-based elements and their key relationships.

Figure 2. Graph representation of the full regional pathway to a BS in EE at the RC
Figure 2. Graph representation of the full regional pathway to a BS in EE at the RC

The new RC can currently demonstrate 10 such operational pathways in five regionally relevant programs of study. Some involve state institutions; others engage universities and colleges outside the state’s educational ecosystem. Because the RC is unable to develop or accredit its own coursework, these external relationships are crucial.

One key workforce driver involves strategic academic partnerships designed to bring strategic economic development to bear in the region. The initial RC KG focuses on the academic side of this equation, which is intended to extend to industrial and governmental program partnerships as well. The academic KG, by itself, has already affirmed some significant insights, not easily recognized without the vivid visualizations made possible by the KG design.

Academic KG Insights

The RC KG clearly shows five key academic relationships and trends.

1. Educational Disciplines

The educational focus of a given region speaks directly to the region’s implicit values. Figure 3 depicts comprehensive regional HS and college concentrations, which are often completed with courses supporting prescribed programs of study and major concentrations of study. These include STEM, business, healthcare, and education, each of which houses programs of study.

Figure 3. Aggregation of educational concentration areas in the region
Figure 3. Aggregation of educational concentration areas in the region

We can see that the region places a high value on STEM education. Given the regional focus on autonomous systems and information technology, this is hardly shocking. The emphasis on business is also not a surprise, as both the military and the federal government are major international buyers in the region. In light of national and regional healthcare demands, the strong emphasis on healthcare doesn’t seem out of place, either. The large concentration of trades, construction, communications, justice, and services reflect levels of workforce diversification, especially for pathways in which specific technical knowledge and skills prevail.

However, the reduced concentration on education programs of study throughout the region is disappointing. Fortunately, the region has a relatively high concentration of trained teachers, which leads to greater quality of instruction. Unfortunately, the pathways to teaching are not as strong as other pathways in the region. This is reflective of the national shortfall of qualified teachers, further fueled by COVID-19 burnout.8  

In each case, the graph database underlying the RC KG captures actual enrollment data, giving a basis for sound quantitative trend analysis. This is most useful in constructing potential pathways that can yield reasonably sized cadres.

2. Certifications

The KG revealed some 133 named certifications related to the concentrations within each field of study. Of these, 38 were unique, career-enhancing certifications awarded by the CC and other educational institutions. The remaining 96 were awarded by recognized professional organizations. Should the notional trend toward stackable certificates underpinning a degree come to full fruition, this data, quantitatively related to the concentrations within the more general fields of study, will prove most useful. These certifications also represent an opportunity for analysis of certification opportunities by concentration to reinforce desirable workforce competencies where workforce training is applicable.

3. Credits

The graph database can capture and visually depict both individual HS courses and the broader concentrations for which external credit may be earned. This includes work-study programs, earned credit programs, and concurrent education offerings. It is significant in that it relates directly to the viability of potential pipelines.

Where offered, such credits provide the incentive to pursue a given program of study. Likewise, where offered, internships may also be represented and depicted. Both appear as valuable components of Figure 2. The left side of Figure 4 shows the available HS credit offerings from which pathways may be derived. These credits represent specific and academically valuable pathway building blocks.

Figure 4. HS college credits and CC articulation agreements as pathway building blocks
Figure 4. HS college credits and CC articulation agreements as pathway building blocks

4. Articulation Agreements

The right side of Figure 4 depicts the articulation agreements that exist between the CC and its partner four-year institutions, including those within the state’s educational ecosystem. The concentrations and their links represent the 10 existing pathways mentioned earlier that are currently available at the new RC. The remaining institutional affiliations show articulation agreements that represent potential pathways available via the regional college and the CC juggernaut. Indeed, enrollment numbers will vary and truly represent regional appetite for a given program of study. Thus, actual enrollments will dictate ultimate pathway viability. Visual representation of the potential pathways, combined with enrollment data, represents a significant point of departure for further academic partnership exploration. 

5. Strategic Variance

The RC exists to support the entire region, but it is essential to appreciate that the three county school districts each is responsive to its county constituencies and localized values. This shows up in how credits are managed and what concentrations are emphasized among the three jurisdictions.

Figure 5 shows the differences between the three regional county school districts, with increasing density from right to left. There is no vertical correlation to individual county school systems. Appreciation of these salient differences is crucial to building a well-balanced set of pathways from which useful cadres may be recruited. Here, the KG shines in defining realities without imposing value judgment.

Figure 5. Volumetric variation among strategic approaches within three regional county school districts
Figure 5. Volumetric variation among strategic approaches within three regional county school districts

Conclusion & Future Direction

The current RC KG provides a useful analytical framework for analysis in pursuit of viable regional pathways. It will reinforce job predictions with a real-world view of potential quantitative readiness to meet demands. It also holds some promise in increasing the probability of building strong cadres by selected programs of study.

However, the KG is but one tool in the academic toolkit, as it is currently far more descriptive than prescriptive. As a work in progress, the KG met with some initial acceptance, but time and further development will determine its ultimate utility in the decision-making process.

Although the academic aspect of the workforce equation is essential to future academic partnerships, this KG cannot stand alone. Industrial and governmental partnerships are equally important in building a balanced regional workforce attuned to its own best interests and needs.

To that end, the next step is to grow the KG to reflect regional industrial and governmental programs extending well beyond internships and apprenticeships that serve to reinforce regional workforce development. This partnership data already exists and, when incorporated, should be instrumental in further identifying both gaps and opportunities to help build industrial and governmental partnerships. Such vital partnerships will serve to grow a vibrant and diversified workforce that truly supports the regional culture.


Kerrigan, Monica Reid, et al. “ATE Regional Centers: CCRC Final Report.” Community College Research Center, Columbia University, May 2007.

Kejriwal, Mayank, Craig A. Knoblock, and Pedro Szekely. Knowledge Graphs: Fundamentals, Techniques, and Applications. MIT Press, 2021.

Ji, Shaoxiong, et al. “A Survey on Knowledge Graphs: Representation, Acquisition, and Applications.IEEE Transactions on Neural Networks and Learning Systems, Vol. 33, No. 2, 26 April 2021.

Antoniazzi, Francesco, and Fabio Viola. “RDF Graph Visualization Tools: A Survey.” Proceedings of the 23rd Conference of Open Innovations Association (FRUCT). IEEE, 27 December 2018.

Dörpinghaus, Jens, and Andreas Stefan. “Knowledge Extraction and Applications Utilizing Context Data in Knowledge Graphs.” Proceedings of the 2019 Federated Conference on Computer Science and Information Systems (FedCSIS). IEEE, 7 October 2019.

Zapata-Carratalá, Carlos, and Xerxes D. Arsiwalla. “An Invitation to Higher Arity Science.” Cornell University, 21 January 2022.

Hogan, Aidan, et al. Knowledge Graphs. Springer, 2021.

Barnes, Adam. “Here’s What’s Driving the Nationwide Teacher Shortage.” The Hill, 21 April 2022.


About The Author
George Hurlburt
George Hurlburt is Chief Scientist at STEMCorp. He serves on the board of advisors for a state regional center and sits on the IEEE IT Professional editorial board as Associate Editor for Departments and Columns. Mr. Hurlburt retired with a Meritorious Civilian Service Award after 38 years as a Navy Senior Systems Analyst, where he pioneered collaborative network architectures for the US Department of Defense and ran a network for its test and… Read More