Living in the information-overloading age, it is difficult to find the right information and identify the resources they need on the websites. As to a user, it is time-consuming in browsing, searching, and making a decision to buy products on online stores. Therefore, many E-commerce websites have implemented recommender systems that intend to provide users with professional recommendation for various types of products and services. Although many recommendation methods have been proposed, there are still some problems like the sparsity and the cold start problems. In addition, some researchers observe there exist users who are biased and products that are controversial. We conjecture that ratings given by biased users or given to controversial products may have impact on estimation accuracy of recommendation. In this thesis, we will examine the measures for user bias and product controversy and propose trust-based-recommendation techniques that take them into account. We evaluate the proposed techniques using the web of trust and rating data collected from the Epinions.com website. It is found that properly setting some parameters, the proposed trust network-based method that incorporates user bias achieve higher recommendation accuracy.