Design of a recommendation system based on the transition graph

A recommendation system has been built for a web resource’s users that applies statistics about user activities to provide recommendations. The purpose of the operation is recommendations in form an orderly sequence HTML pages resource suggested user. ranking procedure uses statistical information transitions between pages. model represented graph; behavior shown as graph by adjacency matrix; transition graph, weighted matrix probabilities vertices constructed. It was taken into consideration may involve entering URL address bar browser or clicking on link current page. user’s finite according determined weight graph’s edges homogeneous Markov chain and considered process random walk with possibility moving vertex. Random Walk Restarts used rank particular Numerical analysis performed actual online store website. initial data sessions are divided training test samples. According sample, probability To assess quality system, accuracy, completeness, Half-life Utility metrics were used. On elements accuracy value 65‒68 % obtained, optimal number list determined. influence parameters systems investigated.