Infuencee Oriented Topic Prediction: Investigating the Effect of Influence on the Author

In this paper, we study the problem of topic adoption prediction for an author within a social academic network. The previous efforts on the problem use topic similarity and topic adoption of co-authors. We model the problem with an influence detection point of view, and propose that the influence on the author is an important factor. Hence, we define a novel influencee prediction based feature. To this aim, in this work, an algorithm is proposed to calculate the influence propagated towards the author. The effect of this feature is explored together with and in comparison to other features used in the literature for the problem. The experiments conducted on Arnet Miner data set show that accumulated influence on author is effective for predicting topic adoption.