One embodiment provides a method for determining credibility of a user in recommending an object. The method comprises monitoring activity relating to an object in an online network, and determining a baseline of activity for the object based on the activity monitored. In response to detecting a user interaction by a first user with the object, information relating to the user interaction is recorded. An analysis of the information recorded is performed in view of information relating to one or more other user interactions by one or more other users with the object. Based on the analysis, a credibility score for the first user is determined. The credibility score represents a degree of confidence in reliability of the first user to recommend the object. Presentation of user recommendations for objects is controlled based on credibility scores to decrease impact of astroturfing, crowdturfing and spamming on the presentation.