Article Preview
Top1. Introduction
The Internet is rapidly developing and the number of various internet resources is increasing nowadays. Daily users deal with various unique web pages with different interfaces. A lot of people spend time in searching the necessary information in the Net, comparing internet-store prices etc. In addition, being on a new page, users often display conservative attitude, which means leaving the page within first minutes due to unwillingness to waste time and feel mental workload of studying new interface. That can be reflected in the reduction of resource audience and hence financial losses for its owners.
This problem can be solved by such web page interface usability optimization that would allow users to avoid discomfort and annoyance. According to this, developers need not to create whatsoever interface but to develop the interface, which would (because of its high usability) attract as many potential clients as possible. Moreover, modern search systems are ranking search results on the basis of evaluating of user behavior which additionally increase high-quality interfaced resources traffic value (Bakaev, 2017). Generally, high-quality interface cannot be created in the first attempt, a qualitative result can be reached only by a number of iterations. At the same time, interface optimization challenges some main difficulties.
Firstly, by reason of unified approach to quantitative measurement of interface usability and concerned with its usage mental workload (Paas & Van Merriënboer, 1993; Alfimtsev & Sakulin & Levanov, 2016; Grabish, 2006) absence, a need of objective function defensible choice for such optimization arises. Secondly, because of big amount if possible parameters and respectively, interface implementation variations (Alfimtsev & Sakulin & Levanov, 2016; Grabish, 2006), corresponding optimization problem will have high dimension. As a result, classical methods are not suitable for web page interface optimization. Thirdly, the people providing this optimizationmust have deep knowledge and experience in web interface development and modification which is often unapproachable. Due to this, a question of the development of web interface optimization methods which would let overcoming described difficulties or reduce their influence on the result becomes actual.
One way to implement such methods is using nature-inspired algorithms since they are preferable to classical ones in high-dimensional problems solution (Yang, 2014; Simon, 2013). The present paper is devoted to usage of such algorithms for web page interface optimization. Further paper has the following structure. In section 2, a review of literature from the fields of interface usability optimization, web page users’ behavior journaling, visualization and analysis is given; works devoted to nature-inspired algorithms usage for web page optimization problems solution are considered; a review of this algorithms and its practical application is given. In section 3 defined method implementation based on applying genetic algorithm (GA), artificial bee colony algorithm (ABC), and charged system search (CSS) algorithm is considered. As the optimization objective we chose interface approximation to the template which was pre-selected by heat maps corresponding to DOM- interface models. In section 4 based on defined algorithms interface optimization experiments are described and the obtained results are given. In section 5 conclusion based on experiments results is given and further research directions are proposed.