Manipulation becomes harder when the manipulators are uncertain about the preferences of sincere voters. Topk elicitation may communicate information about sincere voters' votes to a manipulator, allowing him to vote strategically. In this paper, a multi-round elicitation process, of sincere voters' preferences, is derived that yields to an optimal manipulation with minimal information elicited. The main idea is to study the connection between the amount of information elicited from sincere voters and the probability of manipulation. Through a deep experimental study, this paper answers the question: How many candidates, per sincere voter, are needed to be known for an optimal manipulation? Probabilistic models such as IC and IC-SP are used to complete preference profiles.