Modeling and visualization of media in Arabic

• Mathematical modeling and visualization methods for media in Arabic are developed.
• A new measure, named the Mean Rank Dependency, is introduced, for evaluation of similarity between newspaper issue and a set of its earlier ones.
• Changes in the linguistic content and their connection to fluctuations in societal developments are traced using this measure.
• A new distance function quantifying issues diversity through the Mean Rank Dependency is presented.

In this paper, a novel method for analyzing media in Arabic using new quantitative characteristics is proposed. A sequence of newspaper daily issues is represented as histograms of occurrences of informative terms. The histograms closeness is evaluated via a rank correlation coefficient by treating the terms as ordinal data consistent with their frequencies. A new characteristic is introduced to quantify the relationship of an issue with numerous earlier ones. A newspaper is imaged as a time series of this characteristic values affected by the current social situation. The change points of this process may indicate fluctuations in the social behavior of the corresponding society as is evident from changes in the linguistic content. Moreover, the similarity measure created by means of this characteristic makes it possible to accurately derive the groups of homogeneous issues without any additional information. The methodology is evaluated on sequential issues of an Egyptian newspaper, “Al-Ahraam”, and a Lebanese newspaper, “Al-Akhbaar”. The results exhibit the high ability of the proposed approach to expose changes in the linguistic content and to connect them with changes in the structure of society and the relationships in it. The method can be suitably extended to every alphabetic language media.