Aquisição de conhecimento na knowledge-based view: desenvolvimento de um método de mineração de patentes para apoiar a capacidade absortiva organizacional

Abstract

Knowledge constitutes a fundamental strategic asset of organizations, being decisive for the generation of sustainable competitive advantages, as postulated by the Knowledge-Based View (KBV). From this perspective, innovation emerges as the result of an organization’s ability to acquire, assimilate, transform, and apply knowledge, a process directly associated with the development of absorptive capacity. The search for scientific and technological information, which composes what is known as technical intelligence, finds in patent databases a repository of codified knowledge. However, its exploitation is hindered by the large volume of documents, linguistic heterogeneity, and the terminological complexity characteristic of patent texts. Patent text mining thus constitutes a challenging yet essential task to enable the systematic acquisition of this technical intelligence, acting as an enabling mechanism of absorptive capacity by reducing cognitive, linguistic, and technical barriers to external knowledge. In this context, theoretically guided by the Knowledge-Based View and methodologically grounded in Design Science Research (DSR), this study seeks to answer the following central research question: how can the acquisition of technical intelligence from patents be enabled so as to allow its internalization and articulation with an organization’s internal knowledge through a text-mining method supported by an ontology? The overall objective is to propose a text-mining method, supported by semantic and linguistic ontology, capable of enabling the acquisition, assimilation, and transformation of the technical information contained in patents within the organizational context. The thesis is structured into four interconnected studies that consolidate knowledge on patent mining, map its evolution, and identify gaps, with particular emphasis on the integration between TRIZ and text mining. The research is pioneering in proposing and validating a hybrid patent text-mining method that combines Artificial Intelligence, through large language models, with a derivative logic supported by a functional semantic ontology based on TRIZ physical effects, originally developed for the Portuguese language. The results demonstrate that the articulation between TRIZ, text mining, and AI strengthens organizational absorptive capacity and provides an empirical and conceptual basis for transforming technological information into strategic knowledge, directly supporting Research, Development, and Innovation activities and enhancing the competitive potential of organizations and innovation ecosystems.