Abstract
This doctoral dissertation focused on defining the integration of open innovation and intellectual capital, grounded in Resource-Based Theory, and applied to public university extension in Colombia, so that an artificial-intelligence planning environment could automate the generation of plans for the open-innovation and intellectual-capital processes required to execute extension projects. Recent research indicates that public higher education institutions have not fully defined their role in open-innovation processes and in managing intellectual capital beyond traditional models such as the triple, quadruple, and quintuple helix. These approaches position universities as knowledge generators, yet they do not ensure the systematic transfer of that knowledge to society through university extension. In addition, a substantial portion of the scientific literature highlights major challenges arising from the democratized use of artificial intelligence, which affects the installed capabilities of university extension in light of new social and technological realities. Process models are often designed manually as declarative flow diagrams by experts with deep knowledge of organizational activities. In public university extension, this leads to a large set of activity rules that becomes difficult to maintain due to limited flexibility, scale, and complexity. Therefore, it is necessary to automatically discover the processes required to address social dynamics under uncertainty through university extension, supported by open innovation and intellectual capital. This dissertation incorporates three components. The first integrates intellectual capital with the outbound open-innovation model within the Resource-Based Theory framework to produce a conceptual model that treats intellectual capital as the internal capabilities of public university extension and outbound open innovation as its external capability flows. This approach is developed through a cross-impact matrix and validated in university-extension case studies. The second component addresses process automation through the construction of BPMN 2.0 flow diagrams for education projects, as well as projects involving technological development and consultancy. This provides the baseline knowledge for an artificial-intelligence planning environment based on Hierarchical Task Network (HTN) planning. In this approach, a parsing algorithm translates the BPMN structure into operators, actions, and methods for building the action model (domain). It also enables the construction of planning problems by capturing extension-project information as BPMN flow diagrams. Generating BPMN diagrams is a precondition for building the domain and defining specific HTN planning problems. The third component focuses on plan generation assisted by a pre-planning algorithm using the SHOP2, PyHOP, and GTPyhop planners for implementing or executing the projects described in the BPMN diagrams. Plan generation is supported by HTN planning languages such as PDDL-HTN or HDDL (Hierarchical Domain Definition Language). The experimental results are validated through case studies and consolidated in a platform that integrates these approaches, named PlanProjU (Planning + Project + University). AI Planning addresses the study and computational modeling of capabilities characteristic of human intelligence, especially those related to goal setting, action selection, and decision-making under constraints and changing environments. From the perspective of Artificial Cognitive Systems, this approach seeks to explicitly represent the deliberative reasoning that guides goal-oriented behavior, using symbolic and hierarchical models such as classical planning and HTN planning. These models allow for the decomposition of complex tasks into manageable subtasks and the construction of coherent action sequences. Within this framework, planning is understood as a central mechanism of cognition, as it integrates knowledge, inference, and evaluation of alternatives to anticipate outcomes, optimize resource utilization, and adapt behavior. This makes it a key field for designing artificial systems with cognitive capabilities closer to those of humans. The main contributions of this dissertation can be summarized as follows: (i) the definition of an integrated intellectual-capital and open-innovation model, grounded in Resource-Based Theory and applied so far, for the first time—to university extension, as an alternative to related work that validates conceptual integrations but does not adopt them as an extension-oriented reference framework; and (ii) the proposal of a university-extension model, framed by open innovation and intellectual capital, supported by a hierarchical AI planning environment, together with the construction of that planning environment to automate plans for extension projects. Validation tests were conducted in extension scenarios at Universidad Nacional de Colombia (Manizales campus), Universidad de Caldas, and Universidad Tecnológica de Pereira (UTP). The results showed the need to prepare extension staff to use the BPMN format. For this purpose, PlanProjU includes a BPMN usage guide, and extension experts were encouraged to use the BPMN-GPT tool, which facilitates converting extension PDF documents into BPMN diagrams. The projects followed different formats. In the case of Universidad de Caldas, projects included both formulation-stage documents and final reports from executed projects. In the cases of Universidad Nacional de Colombia (Manizales campus) and Universidad Tecnológica de Pereira, the institutions shared executed projects in formulation documents. In total, 36 tests were performed on the PlanProjU platform using these projects. After reviewing the planning domain and problem, the extension user can approve or reject the generated plans. At the end of the validation stage, PlanProjU executed without issues. Moreover, in the instrument administered to extension staff, respondents stated that the platform is user-friendly and produces plans that can be implemented within the organization. Finally, based on the concepts and the proposed model, an experimental web platform for open innovation and intellectual capital in university extension was designed and implemented. Its main function is to enable project formulators to upload a BPMN diagram of their project in advance and automatically generate the planning domain, while allowing the definition of planning problems so that SHOP2, PyHOP, and GTPyhop can produce plans for project execution or implementation automatically. The system was designed for extension projects in public universities in Colombia under a client–server architecture, using free and cross-platform tools so that it can be accessed by interested individuals or institutions.