On approximation of bookmark assignments
MFCS'07 Proceedings of the 32nd international conference on Mathematical Foundations of Computer Science
Optimal approximability of bookmark assignments
Discrete Applied Mathematics
| Hi-index | 0.00 |
In a Web site, each page v has a certain probability pv of being requested by a user. The access cost of page v, is c(v) = pv · c(r, v), where c(r, v) is the cost of the shortest path between the home page, r, and page v. The access cost of a Web site is the sum of the access cost of all its pages. The cost of a path is measured in two ways. One measure is in terms of its length, where the cost of the path is simply the number of hyperlinks in it. The other measure is in terms of the data transfer generated for traversing the path. This research work concerns the problem of minimizing the access cost of a Web site by adding hotlinks over its underlying structure. We propose an improvement on Web site access by making the most popular pages more accessible to users. We do this by assigning hotlinks to the existing structure of the Web site. The problem of finding an optimal assignment of hotlinks is known as the hotlink problem.Consider a directed graph G = (V, E), where V represents the pages of a Web site connected via hyperlinks E. We prove that the hotlink assignment problem is NP-hard, and give some theoretical results related to this problem. We present heuristic algorithms which are tested and compared by simulation on real and random Web sites. We develop The Hotlink Optimizer (HotOpt), a new software tool that finds an assignment of hotlinks reducing the average number of steps in a Web site. HotOpt is empowered by one of the algorithms presented in this dissertation.