The prediction of protein secondary structures is not only of great importance for many biological applications but also regarded as an important stepping stone for solving the mystery of how amino acid sequences fold into tertiary structures. Recent research on secondary structure prediction is mainly based on widely known machine learning techniques, such as Artificial Neural Networks and Support Vector Machines. The most significant breakthroughs were the incorporation of new biological information into an efficient prediction model and the development of new models which can efficiently exploit suitable information from its primary sequence. Hence this paper reviews the theoretical and experimental literature of these models with a focus on informational issues involving evolutionary and long-range information of protein sequences. Furthermore, we investigate several key issues in protein data processing which involve dimensionality reduction and encoding schemes.