Networking via the Internet is becoming very popular. Many people meet up in online social networks where they form friendships, share the same interests and discuss several topics. By interacting with each other, people influence one another's opinions. Analyzing opinion formation in online social networks yields significant information for market research. This paper introduces mining services which explain and predict opinion formation of the network members on the basis of their social relationships by combining methods from text mining and swarm intelligence. The approach is illustrated by taking an online gaming community as an example.