Reacting to a published post on a social media is one of the main activities of users which can happen in different forms comprising to like the post, leave a comment or reshare it. Finding a way to predict the size of users future interactions and more interestingly identifying the users who are going to react to a post are the two important research topics which benefit different domains from efficient advertising campaign to enhanced content delivery systems. In this paper, we aim to predict the users who are going to react to a newly published post in future. Toward this aim, we implement a novel approach based on Point-wise Mutual Information (PMI) which derives users latent similarities from their interactions log and exploits them to predict future interacting users. The proposed method is evaluated using a large dataset of Flickr including 2.3M users and 11.2M published photos. The empirical findings support the idea of employing interactions log to detect future likers of posts by achieving noticeable prediction results for the tested dataset. Moreover, the analysis of the prediction task implies that likers prediction for the photos of publishers with a high number of followers and engagements is more accurate than the other publishers photos