The high volume of microblogs produced daily together with their rich social structure makes microblogs' better query and filtering ever challenging. In the literature, most of the existing ranking methods are based on the overall popularity of the authors and the tweets without considering author's expertise. In this paper, we propose the topical authority-based ranking methods for social networks like Twitter and investigate how the underlying topical feature modeling can be optimized for performance boosting. In particular, we present a detailed study on the empirical distributions of the topical features. We propose the use of specific parametric forms for different features, which we believe to be crucial as the value of the cumulative distribution function is explicitly used for topical authority ranking. We applied the extended topical authority-based ranking method to a Twitter dataset for ranking keyword-matched microblogs. The experimental results show that our proposed approach outperforms a number of existing approaches by a large margin which verify the effectiveness of our proposed features and the importance of the topical authority for ranking microblogs.