High-quality ontology engineering traditionally prioritizes complete, reusable domain models. While efective for broad reuse, this “ontology-first” approach can misalign with the needs of strategic decision makers, who need targeted, actionable insights on constrained timelines. This paper introduces a value-first framework that inverts this process, beginning with the strategic goals, jobs, and knowledge gaps of business leaders to generate lean, purpose-built knowledge graph that delivers immediate value. In a pilot project with CompanyA, we applied this framework to the wind energy ecosystem, successfully answering 15 distinct knowledge questions. To demonstrate this, we focus on one such question out of 15, analyzing data from 35 companies collected at WindEnergy Hamburg 2024. Our findings show that this approach not only answers knowledge questions efectively through tailored visualizations but also uncovers critical blind spots-such as the intermediary roles of consulting firms-that conventional business ecosystem analyses would necessarily miss.
A business ecosystem is not just a collection of isolated entities but a dynamic system, much like a
biological one [
Therefore, overlooking relationships or the implicit roles of certain actors within these
ecosystems [
Current OKGE methodologies, however, create a fundamental mismatch with the needs of
strategic decision-makers (e.g., Business Development Managers, Chief Innovation Oficers). For example,
established OKGE frameworks (e.g., METHONTOLOGY [
This mismatch represents a fundamental quality and expectation gap in how OKGE creates value.
Current methodologies treat strategic insight as an afterthought—an emergent property of a technically
sound model. The philosophy is: build a complete, consistent ontology first, assuming strategic value
will eventually follow. Strategic decision-makers need the inverse: a targeted ontology, designed
specifically around their knowledge gaps, to help them get their job done [
Therefore, bridging this gap requires a paradigm shift. Ontology engineering must explicitly organize
the development process around strategic business objectives as the primary design driver, rather than
treating strategic value as an emergent consideration. This demands new frameworks that explicitly
translate strategic goals directly into ontology design decisions— from initial value definition to final
value delivery—to get decision makers’ job done [
This pressing pragmatic need leads to our research question: How can strategic business objectives be systematically operationalized as the foundational organizing principle for ontology and knowledge graph engineering to enable competitive intelligence in dynamic business ecosystems?
To address this, we introduce BEAR [
As we have argued in the introduction, analyzing business ecosystems requires bridging two distinct research domains: high-level business strategy with ontology and knowledge graph engineering (OKGE). This section builds the arguments for our approach in two steps. First, we review the state of business ecosystem literature, to identify a critical gap: the lack of methods that are both semantically rich and empirically grounded. Second, we argue that while OKGE is perfectly suited to fill this gap, current ontology-centric methodologies are not explicitly designed for strategic analysis. Therefore, the following review will show a consistent pattern of treating strategic insight as a secondary, emergent property rather than the primary design driver, thereby establishing the critical knowledge gap that the BEAR framework is designed to fill.
To analyze business ecosystems, recent data-driven methodologies utilize natural language processing
and text mining on large unstructured text corpora (e.g., company reports) [
Beyond data-driven approaches, structured conceptual modeling ofers alternative ways to analyze
business ecosystems, focusing on intensions [
Established OKGE frameworks, such as METHONTOLOGY, ofer robust methods for domain analysis
but are fundamentally product-centric [
This product-centric focus persists even in collaborative, eficiency-driven methodologies with
disintermediating eforts of ontology engineers [ 21]. UPON lite, which is a light version of the UPON
methodology [22], rightly critiques traditional ontology building (e.g., [
The Extreme Design (XD) methodology is notable for explicitly incorporating business value into its design philosophy, inspired by Extreme Programming practices [24]. Its extreme lightweight ontology, maintenance, and prototyping principles aim to deliver quick business value, rather than building for the abstract future [24]. XD tries to capture this value through planning, where customers define their needs via desired features and CQs. However, this is where the methodology’s business value philosophy disconnects from its product-centric process. While customers are asked to define “business value”, the mechanism for this is still the CQ, a tool for specifying the product’s features (See Table 1). Furthermore, XD does not explicitly describe how to derive the initial “baseline ontology” [25] from strategic goals. Ultimately, despite its aims and starting point, the process later focuses on building the ontology itself, rather than building the ontology around the given strategic goals and knowledge gaps of the stakeholders.
Even highly innovative structural approaches, such as the Modular Ontology Modelling methodology (MOMo), illustrate this focus on the ontology as the end product. Building upon the Extreme Design Methodology, MOMo’s guiding principle is not the traditional taxonomical (is-a ) hierarchy. Instead, it prioritizes the modularity of the ontology, viewing each module as a part of the whole [26]. Despite this novel and innovative approach to ontology engineering, the first two steps, as well as the example use case descriptions of the MOMo workflow, indicate that the methodology still focuses on the ontology itself, seeing it as the end product [26]. This journey through the current landscape, from traditional to recent, reveals a clear and consistent theme. If we generalize, they ask the question, “What must this ontology be able to represent?” This inevitably forces the strategic decision maker to adapt to the model’s structure.
Therefore, a critical gap remains for a methodology that inverts this process. We argue for a shift away from defining a model’s features and toward delivering its value. In this context, we define value not as a model’s technical completeness or reusability but as the degree to which it closes a specific, strategic knowledge gap for a decision-maker to get their job done and reach their or their organization’s business goals. Therefore, an ontology became a means to an end rather than an end in itself. These arguments lead us to ask diferent type of question: “What strategic question must this ontology answer to get the job of decision maker done?”This shift is the foundation of the BEAR framework, which we introduce next.
The BEAR framework has three core processes: Value Definition, Creation, and Delivery. This section describes the principles and workflow of each process in a sequence. The next section presents a detailed case study to provide a clear illustration of this framework in action.
BEAR’s value definition process begins by treating decision makers as customers. The first process is
to identify their high-level business goal—goals that are strategic, informal objectives they express in
natural language. From these goals, BEAR defines the stakeholder’s specific “Jobs To Be Done (JTBD)”—
a progress a customer is trying to make in a given circumstance to advance their goals [
Through collaborative elicitation (e.g., workshops, interviews, meetings), we find these business goals, jobs, and knowledge gaps, subsequently formulating the gaps into single, consensus-driven knowledge questions (KQ). Following iteratively refining the KQ with decision makers, BEAR engineers analyze it to identify the core Representational Units (RUs), which are the minimal semantic components (classes, properties, relationships), in our context, needed to answer the KQ [28]. These RUs are then formalized into a lean Seed Ontology (SO) similar to baseline ontology [24, 25] in OWL [29], leveraging its expressive power and formal semantics.
This SO is neither a comprehensive domain nor an application ontology. Instead, it formally represents what the stakeholder needs to know to get their job done. It functions as a lightweight, reusable pattern for value-first modeling, emphasizing pragmatic suficiency over exhaustive domain coverage, much like ontology design patterns (ODPs) promote reuse and modularity in ontology engineering [26]. In a business context, this SO provides a reusable blueprint that ontology engineers can "hire" to address similar knowledge gaps across diferent ecosystems, ensuring alignment with stakeholder jobs-to-be-done [27].
In the value creation process, SO becomes the template for designing the data collection methodology. It ensures we only target the evidence that addresses the stakeholder knowledge gap. Based on the SO and stakeholder agreement, we specify the exact data sources (e.g., expos, websites, databases, social media) and collection protocols (e.g., interviews, surveys, web scraping) for the value creation process.
Once we collect the data, BEAR analyzes RUs within the collected evidence and iteratively maps them to the developing SO. During this mapping process, BEAR employs a form of reification [ 30] to handle evidence where specific information about an entity is ambiguous or incomplete (e.g., an entity is mentioned by company type but not by company name).
This iterative mapping evolves the lean SO into a more comprehensive application ontology, and we argue, potentially towards a reference ontology through continued iterations. Crucially, deductive reasoning, employing standard OWL reasoners [29], plays a significant role in consistency checking through this SO evolution.
BEAR’s third and final process, value delivery, answers the stakeholder’s KQ. Following the business
ecosystem literature, we develop interactive, tailored visualizations [
Importantly, these inferences are not mere technical artifacts. They are the engine for revealing strategic insights that decision makers are looking for and exposing critical blind spots— the core value BEAR was designed to deliver.
To deliver these insights efectively, BEAR employs and advocates custom, flexible visualizations built with libraries like D3.js [31]. We argue that standard of-the-shelf ontology visualization tools are too rigid to answer specific knowledge questions, as they are often designed to display the overall schema, rather than enable exploration of instance-class level interactions [32, 33]. In contrast, a tailored visualization can provide more than just a picture of the data; it is a critical mechanism for delivering value— an answer engineered to close the knowledge gap to get the stakeholder’s job done to reach business goals.
To validate and illustrate the BEAR framework, we applied it in a pilot project with CompanyA to a real-world strategic challenge within the wind energy business ecosystem. The project followed the three core processes of the BEAR framework.
Because we advocate for open science and reproducibility, we have made the research artifacts from
this pilot publicly available in our GitHub repository [
The value definition process began with a collaborative workshop with senior leaders at CompanyA. Their high-level business goal was clear: to drive revenue growth in the increasingly crowded wind energy ecosystem. Nevertheless, they were struggling to reach that goal. Traditional market analysis and business development eforts felt like they were competing against the luck [ 27]—trying to innovate their business models, often with no real way to predict their organization’s success.
At this moment of struggle, we helped them articulate the specific jobs to be done, which prevented them from making progress as advocated in the jobs to be done theory [27]. The real job they needed to get done was: “Help us see the hidden relationships of our ecosystem so we can confidently identify new, high-margin service opportunities and stop wasting resources and company capabilities on lowprobability bets.” This job was not the only one we identified; however, for this paper, we focus on this one. Therefore, within this context, getting this job done was one of the ways to achieve their broader business goal of revenue growth.
Beginning to get this job done, we had to move from their abstract struggle to concrete knowledge gaps. Here we argue that, to get this job done, they also might need human resources, special equipment, and many other resources; however, another important point relevant to the ontology engineering is the knowledge gap they had: “company positions within wind energy ecosystem achieved through product/service delivery interactions”.
As competency questions bridges the gap between the ontology engineers and stakeholders [
Analyzing this KQ revealed related Representational Units (RUs), such as “Company”, “Product Delivery Interaction”, and “Service Delivery Interaction”. A systematic analysis was used to distill the KQ into its core semantic components (Table 2). We then formalized these RUs in OWL2 to engineer the Seed Ontology (SO), with design decisions guided explicitly by the Value-First Principle (Table 3).
For this application, we used the SO to guide the data collection methodology creation process directly.
We designed a semi-structured survey for rapid, open-ended data acquisition at industrial expos [
As anticipated, this iterative modeling process uncovered new, more abstract classes not present in our initial SO. For example, we created a wbeo:Operator parent class to unify wbeo:GridOperator and wbeo:WindTurbineOperator (See Figure 2). To manage these emergent abstractions with ontological rigor, we applied the established principle of single inheritance [28]. This process also forced us to handle incomplete data, for which we used reification—creating typed blank nodes for relationship modelling (Figure 3).
Consequently, through this iterative process of mapping and refinement, the SO evolved into the
Wind Business Ecosystem Ontology (WBEO) [
In our wind energy application, we first modeled two SPARQL queries [
Finally, we exported these results as a JSON file and fed them into an interactive visualization developed with D3 [31]. This final tool did more than just displaying the inferred data, it answered the stakeholder’s KQ directly, revealing strategic blind spots through features like filtering and granularity adjustments (See Figure 4).
In conclusion, the project resulted in several meetings of the interactive visualization with key decision-makers within diferent departments at CompanyA. During the meetings, stakeholders could explore the interactive visualization and ask follow-up questions, which led to further insights and discussions about their business ecosystem. The value was directly apparent, leading to two key outcomes: a richer, shared understanding of their business ecosystem and the formal approval of a new pilot project. This second project will test the BEAR framework’s utility in a new business context, afirming its role as a repeatable and valuable strategic tool not just for business ecosystem analysis, but also for other strategic decision-making contexts.
Uncovering complex structural relationships and strategic blind spots within business ecosystems
demands a semantic approach, not just syntactic analysis. While existing OKGE methodologies are
inherently semantic, they do not explicitly organize and focus their engineering processes based on
business goals, their jobs, and knowledge gaps of the stakeholders [
To answer these KQs efectively, BEAR’s robust handling of incomplete data—an everyday reality in
business ecosystems yet unaddressed in literature [
Our seed ontology approach presents a fundamental tension in ontology engineering. By designing minimal ontologies for specific KQs, we achieve direct value delivery where traditional methodologies struggle. Yet, this focus risks creating “conceptual silos”—isolated ontologies that answer one question with pragmatic suficiency, but do not communicate.
Our solution to this paradox lies in BEAR’s relationship to modular ontology engineering philosophy.
Our seed ontologies are inherently modular, not in their structure, but in their value, a diferent sense
of MOMo [26]. Each seed ontology is a self-contained “value module”, that delivers specific insight. The
architectural challenge is, however, to prevent these modules from becoming disconnected. How do
we prevent diferent knowledge questions from leading to disconnected seed ontologies? For example,
in our pilot, we answered in total of 15 distinct KQs, and while we could essentially merge them
into a single Wind Business Ecosystem Ontology (WBEO) [
Our preliminary answer lies in treating seed ontologies as specialized modules in an evolving reference ontology. Evidently, the WBEO began from a single KQ; however, with iterative refinement—each new data point, each emergent class (like wbeo:Operator abstraction)—it moved closer to a comprehensive domain model (See Figure 2). This suggests a clear development pathway: seed ontologies for immediate value, which iteratively build and enrich a larger reference ontology for a long-term knowledge asset.
However, we argue that to achieve this, we must integrate an upper ontology (e.g., Basic Formal Ontology [28]) to map each seed ontology to a common framework.
A value-first paradigm demands value-first metrics. While foundational for model soundness, traditional
metrics like technical correctness and completeness measure the engineering artifact, not its business
impact. They can tell us if a model is built correctly (e.g., by validation through competency questions [
We argue that, to bridge this gap, a project’s success must be judged by its disposition to fill the stakeholders’ knowledge gap. Based on the insights from our pilot project, we propose an initial set of value-first quality metrics (Table 4). This initial set permits further research and systematic validation.
Our framework, while promising, evidently has apparent limitations that define our future work. The current implementation relies on manual data collection and RU analysis, a critical scalability concern raised by our pilot stakeholders, which is our most explicit challenge. While our semi-structured survey is a step toward rapid, ontology-aligned data acquisition at events like industrial expos, true scalability requires automation. Therefore, we have initiated a new pilot focused on automated blind spot detection using graph pattern recognition. We are also exploring the use of LLMs to semi-automate RU extraction
Does the final answer enable a specific, Bridges the gap between data and exinformed decision or action? ecution; the output must be usable.
Is the developed ontology suficiently lean to deliver the maximum insight required by the knowledge question?
Prevents over-engineering and focuses resources on value, not exhaustive modeling.
How quickly does the engineering process move from business goal to actionable insight?
Measures the agility of the framework, which is critical for decision-makers on tight timelines.
To what degree was the stakeholder’s The ultimate success metric, rooted in
initial knowledge gap resolved? the Jobs-To-Be-Done framework [
Another important aspect is the philosophical grounding of the BEAR framework. Our implicit commitment to scientific and ontological realism [ 40, 41], unlike cognitive [42] or linguistic [43], which aligns with the Basic Formal Ontology (BFO) [28], must be formalized to provide a more solid grounding for future development and integration.
Finally, while our single case study demonstrates feasibility and the successful handling of 15 distinct KQs suggests adaptability, generalizability must be validated. Even though it is due to its focus on one ecosystem (wind energy) and a subset of data (37 surveys), we have already begun a new pilot project in a diferent business context to test BEAR’s applicability across diverse contexts and stakeholders, and we will share these results in future work.
As a conclusion, the complexity of modern business ecosystems demands semantic analysis—this much has been shown in our wind energy study. However, semantic capability alone is insuficient. Strategic decision makers do not need ontologies; they need answers. BEAR demonstrated this by inverting the traditional engineering process—starting with the answer they needed rather than the model required—we can deliver immediate strategic value while building toward a comprehensive knowledge infrastructure. The question is no longer whether ontologies can support strategic decision making, but whether we are willing to reorganize our engineering practices around the value they must deliver.
The financial support by the Austrian Federal Ministry of Digital and Economic Afairs, the National Foundation for Research, Technology, and Development, and the Christian Doppler Research Association is gratefully acknowledged.
The authors used Grammarly for grammar and spelling support. The final manuscript was reviewed and edited by the authors, who are fully responsible for its content. systems 16 (2001) 26–34. [21] A. De Nicola, M. Missikof, A lightweight methodology for rapid ontology engineering, Communications of the ACM 59 (2016) 79–86. [22] A. De Nicola, M. Missikof, R. Navigli, A proposal for a unified process for ontology building: Upon, in: International conference on database and expert systems applications, Springer, 2005, pp. 655–664. [23] M. Poveda-Villalón, A. Fernández-Izquierdo, M. Fernández-López, R. García-Castro, Lot: An industrial oriented ontology engineering framework, Engineering Applications of Artificial Intelligence 111 (2022) 104755. [24] M. Hristozova, L. Sterling, An extreme method for developing lightweight ontologies, in: Workshop on Ontologies in Agent Systems, 1st International Joint Conference on Autonomous Agents and Multi-Agent Systems, 2002. [25] M. Boicu, G. Tecuci, D. Marcu, Ontologies and the Knowledge Acquisition Bottleneck, in: Proceedings of the IJCAI-01 Workshop on Ontologies and Information Sharing, Seattle, WA, USA, 2001, pp. 12–18. [26] C. Shimizu, K. Hammar, P. Hitzler, Modular ontology modeling, Semantic Web 14 (2023) 459–489. [27] C. Christensen, T. Hall, K. Dillon, D. S. Duncan, Competing against luck, The story of innovation and customer choice. First edition. New York, NY: HarperBusiness an imprint of HarperCollins Publishers (2016). [28] R. Arp, B. Smith, A. D. Spear, Building ontologies with basic formal ontology, Mit Press, 2015. [29] P. Hitzler, M. Krötzsch, B. Parsia, P. F. Patel-Schneider, S. Rudolph, et al., Owl 2 web ontology language primer, W3C recommendation 27 (2009) 123. [30] M. Giunti, G. Sergioli, G. Vivanet, S. Pinna, Representing n-ary relations in the semantic web,
Logic Journal of the IGPL 29 (2021) 697–717. [31] M. Bostock, D3, https://github.com/d3/d3, 2025. [Online; accessed 2025-07-10]. [32] S. Lohmann, V. Link, E. Marbach, S. Negru, Webvowl: Web-based visualization of ontologies, in: International Conference on Knowledge Engineering and Knowledge Management, Springer, 2014, pp. 154–158. [33] Graphdb, https://graphdb.ontotext.com/documentation, 2025. [Online; accessed 2025-08-02]. [34] M. Uschold, M. Gruninger, Ontologies: principles, methods and applications, The Knowledge
Engineering Review 11 (1996) 93–136. doi:10.1017/S0269888900007797. [35] WindEnergy Hamburg, Global on & ofshore event, 2024. URL: https://www.windenergyhamburg.
de/, [Online; accessed 2025-05-10]. [36] D. Küçük, D. Küçük, Ontowind: An improved and extended wind energy ontology, arXiv preprint arXiv:1803.02808 (2018). [37] D. Vrandečić, S. Pinto, C. Tempich, Y. Sure, The diligent knowledge processes, Journal of Knowledge
Management 9 (2005) 85–96. [38] A. Gómez-Pérez, M. C. Suárez-Figueroa, Neon methodology for building ontology networks: a scenario-based methodology (2009). [39] C. Shimizu, P. Hitzler, Accelerating knowledge graph and ontology engineering with large language models, Journal of Web Semantics (2025) 100862. [40] A. Chakravartty, Scientific Realism, in: E. N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy, summer 2017 ed., Metaphysics Research Lab, Stanford University, 2017. URL: https://plato.stanford. edu/archives/sum2017/entries/scientific-realism/. [41] B. Smith, W. Ceusters, Ontological realism: A methodology for coordinated evolution of scientific ontologies, Applied ontology 5 (2010) 139–188. [42] G. Guizzardi, A. Botti Benevides, C. M. Fonseca, D. Porello, J. P. A. Almeida, T. Prince Sales, Ufo:
Unified foundational ontology, Applied ontology 17 (2022) 167–210. [43] A. Gangemi, N. Guarino, C. Masolo, A. Oltramari, L. Schneider, Sweetening ontologies with dolce, Springer, 2002, pp. 166–181.