In recent times, Online Social Networks have been predominantly used by commercial houses for promoting their products. The key approach in this context is to choose a limited number of highly influential users such that if they become the initial adopter of the product then the influence (and hence profit) in the network gets maximized. Existing studies on this problem consider that all the links of the network contain positive influence probabilities only. However, in practice, a social network contains links for both positive as well as negative probabilities, and such networks are called Signed Social Networks. In this paper, we study the profit maximization problem in the context of signed social networks under the popular Polarity Linked Information Diffusion (PLID) Model. We show that, in our case, the profit function is non-negative, non-monotone, and non-submodular. We propose two solution approaches for this problem. The first one is an iterative greedy approach, which works based on marginal gain computation, and the other one is a local search-based approach, which iteratively improves the quality of the solution. Both methods have been analyzed to understand their time and space complexity. A number of experiments have been carried out to evaluate the effectiveness and efficiency of the proposed solution approaches. We observe that in most of the instances the seed set selected by the proposed solution approaches leads to more profit compared to the existing methods.