The growth in the utilization of text mining tools and techniques in the last decade has been primarily driven by the increase in the sheer volume of unstructured texts and the need to extract useful and more importantly, quality information from them. The impetus to analyse unstructured data efficiently and effectively as part of the decision making processes within an organization has further motivated the need to better understand how to use text mining tools and techniques. This paper describes a case study of a stock spam e-mail architecture that demonstrates the process of refining linguistic resources to extract relevant, high quality information including stock profile, financial key words, stock and company news (positive/negative), and compound phrases from stock spam e-mails. The context of such a study is to identify high quality information patterns that can be used to support relevant authorities in detecting and analyzing fraudulent activities.