In this article we present the matchmaking problem in industrial symbiosis where wastes from one company is matched with resources of another company that could be substituted. Identifying potential matches is difficult, as it is based on process-specific knowledge that certain wastes can be used for specific processes. Capturing this knowledge in waste-resource matching rules manually is time-consuming. Therefore, we argue that a Natural Language Processing (NLP)-based approach of semi-automatically extracting rules from domain-specific data sets could be a viable approach to solving this problem. The basic NLP problem to solve is to find similar concepts (synonyms), part-whole relationships (meronyms), and " is a " relationships (hyponyms). Synonyms are important for finding wastes and resources that are named differently but refer to the object. Meronyms are part-whole relationships that may help to identify wastes with components that could be used as a resource. Hyponyms allow for building taxonomies. We present the results of an initial literature survey of algorithms that are able to find these relationships in large sets of unstructured text documents. Furthermore, we propose a research approach for further extending the literature survey and testing the existing algorithms on small test cases and realistic matchmaking case. For future work, additional problems that fall into the NLP category can be addressed such as semi-automatically identifying processes for converting wastes into resources.