Simple models of the content duration and the popularity of television content

Abstract

This paper presents parametric statistical models for the content duration (running length) and the popularity of TV programs. The parameters of the models are obtained by curve fitting data obtained from the Internet Movie DataBase (IMDB). Later, the accuracy of the models is numerically assessed, providing an estimation of the error, together with a use case example. The content duration model has been obtained exclusively from the actual duration of TV programs, which is intrinsic to the content, instead of the file size of encoded videos which is application dependent. It is also worth emphasizing that, unlike several previous studies, the model does not account for User Generated Content. The model can be used to obtain the file sizes of video items in video delivery services for different formats or resolutions, by applying their characteristic video encoding rates. On the other hand, this paper shows parametric popularity models adjusted to fit the distribution of user ratings in the IMDB database. The popularity models can be used to estimate the probability of access of programs in video on demand services. Therefore, the content duration and popularity models hereby presented are useful to evaluate the performance of video delivery services dealing specifically with TV content. (C) 2013 Elsevier Ltd. All rights reserved.