The PLATOS project aims at investigating stance and argument detection on automatically extracted topics in social media text by combining linguistic knowledge and machine learning. In the proposed research, we want to build further on research in aspect-based sentiment analysis for the automatic extraction of topics, their related features and sentiment and the stance taken by the author. In addition, we will push the state-of-the- art in the research field of argumentation mining by modeling argument structures in social media text in order to detect why people have a certain viewpoint about a concerned issue. The proposed research will be evaluated for the use case of political social media text, which is a very relevant test domain because (1) it is important to have a flexible approach to detect emerging and trending political topics, (2) political posts have by nature a more explicit stance and persuasive goal, which makes them a good case for argumentation research and (3) political UGC shows a large variety of language use, ranging from very noisy (blogs, comments) to more standard text (political posts from representatives).