Process modeling requires both data (chemical reaction yields, kinetic constants, cost estimates, environmental indicators, etc.) and knowledge (operation models and formulations, alternative processes and technologies, etc.). Searching in databases and published research may provide such information, but there is a lack of systematic methods and tools guiding this procedure. The present work describes and assesses an information retrieval methodology that is part of a proposed retrieval and extraction cycle addressing this problem. Two query construction methods for sampling academic databases are proposed, assessed and compared. Departing from a seed corpus of a limited number of papers, Scopus® is used as an academic database to retrieve literature containing information associated with pyrolysis processes of waste plastic. It is found that, with minimal human intervention, the methodology is able to return a ranked list of candidate documents that have a considerable (linguistic) relevance.