Smart Internet Search with Random Neural Networks

Abstract

Web services that are free of charge to users typically offer access to online information based on some form of economic interest of the web service itself. Advertisers who put the information on the web will make a payment to the search services based on the clicks that their advertisements receive. Thus, end users cannot know that the results they obtain from web search engines are exhaustive, or that they actually respond to their needs. To fill the gap between user needs and the information presented to them on the web, Intelligent Search Assistants have been proposed to act at the interface between users and search engines to present data to users in a manner that reflects their actual needs or their observed or stated preferences. This paper presents an Intelligent Internet Search Assistant based on the Random Neural Network that tracks the user’s preferences and makes a selection on the output of one or more search engines using the preferences that it has learned. We also introduce a ‘relevance metric’ to compare the performance of our Intelligent Internet Search Assistant against a few search engines, showing that it provides better performance.