Dynamic layout optimization for newspaper web sites using a controlled annealed genetic algorithm

Abstract

A newspaper layout mainly consists of rectangles laid out on a surface in a way that produce no gaps, and looks good. The pagination problem for newspaper web sites is trying to make these layouts automatically. We present an annealed genetic algorithm with article control functions to calculate the rectangle position in order to solve the pagination problem. Articles contain metadata that describes individual design requests, and preferred position for each article in categories headliner, float article and priority article. The positioning request tell the algorithm which articles to position first, and which to position around the other to produce a good fit. A fitness function is used to measure good layouts. The method is tested on two groups containing respectively 10 and 6 participants. One group tests the fitness function performance with the control function implemented and the other group tests the prototype which presents a publishing tool with a dynamic layout generator. The prototype test group contains experienced persons working in large Norwegian newspaper web sites. Although we wanted to add control to the automation there were common agreement from the participants that a professional newspaper web site would need a higher level of control. There was also a common agreement that other areas within web site publishing could benefit a layout generator.