Preference queries, in general, retrieve most preferred tuple(s) in a given relation r(R) for a user-specified preference formula in either relational algebra or SQL. Due to the huge volume of data in today's databases and world-wide-web information systems and peculiar characteristics of preference formula defined in terms of r(R) x r(R), processing preference queries could be time-consuming, and some techniques for improving the process of evaluating preference queries are needed. This paper will address some semantic techniques and heuristics to improve the processing of preference queries. The technique will explore semantic relationship among a set of attributes in a preference formula and utilize this kind of semantic information to evaluate preference queries better.