Complex systems (such as automation systems), from here on referred as “products” are usually difficult for customers to specify since there are a lot of parameters to be defined and the customer should perfectly know what is really needed and important for the supplier in order to provide for a proper system. As a result, creating forms and templates for the customer request specification entry helps only for relatively simple tasks and the completely digital request acquisition and processing is still a matter of future work. Currently, the original request specification comes from the customer in various ways (texts, images, diagrams, a phone or a direct talk to the company’s sales representative) and the results of the analysis of this specification are often forwarded further to the back-office in a form of free or semi-structured text written in natural language. Since this text is the main source of information about the customer request, it is very important to extract as much information from it as possible. The paper reports the research and development work on semantic text analysis for information extraction from customer requests written in natural language. The core of the work is development of methods for finding a pre-defined list of terms (product parameters that are important for the order specification) in a fuzzy (similarity-based) manner with the help of synonym dictionaries. The results are illustrated on a case study from the automation equipment producer Festo SE & Co KG.