This paper focuses on preserving the privacy of sensitive patterns when inducing decision trees. Our record augmentation approach for hiding sensitive classification rules in binary datasets is preferred over other heuristic solutions like output perturbation or cryptographic techniques since the raw data itself is readily available for public use. We describe the process and an indicative experiment using a prototype hiding tool.

Abstract—Most explanations of the user behavior while interacting with the web are based on a top-down approach, where the entire Web, viewed as a vast collection of pages and interconnection links, is used to predict how the users interact with it. A prominent example of this approach is the random-surfer model, the core ingredient behind Google’s PageRank. This model exploits the linking structure of the Web to estimate the percentage of web surfers viewing any given page. Contrary to the top-down approach, a bottom-up approach starts from the user and incrementally builds the dynamics of the web as the result of the users ’ interaction with it. The second approach has not being widely investigated, although there are numerous advantages over the top-down approach regarding (at least) personalization and decentralization of the required infrastructure for web tools. In this paper, we propose a bottom-up approach to study the web dynamics based on web-related data browsed, collected, tagged, and semi-organized by end users. Our approach has been materialized into a hybrid bottom-up search engine that produces search results based solely on user provided web-related data and their sharing among users. We conduct an extensive experimental study to demonstrate the qualitative and quantitative characteristics of user generated web-related data, their strength, and weaknesses as well as to compare the search results of our bottom-up search engine with those of a traditional one. Our study shows that a bottom-up search engine starts from a core consisting of the most interesting part of the Web (according to user opinions) and incrementally (and measurably) improves its ranking, coverage, and accuracy. Finally, we discuss how our approach can be integrated with PageRank, resulting in a new page ranking algorithm that can uniquely combine link analysis with users ’ preferences. Index Terms—Collaborative search engine, personalization, person-oriented system, user information space.

For a place that gathers millions of people the Web seems pretty lonely at times. This is mainly due to the current predominant browsing scenario; that of an individual participating in an autonomous surfing session. We believe that people should be seen as an integral part of the browsing and searching activity towards a concept known as social navigation. In this work, we extend the typical Web browserpsilas functionality so as to raise awareness of other people having similar Web surfing goals at the current moment. We further present features and algorithms that facilitate online communication and collaboration towards common searching targets. The utility of our system is established by experimental studies. The extentions we present can be easily adopted in a typical Web browser.

We use genetic algorithms to evolve classification decision trees. The performance of the system is measured on a set of standard discretized concept learning problems and compared (very favorably) with the performance of two known algorithms (C4.5, OneR).

This work deals with stability in incremental induction of decision trees. Stability problems arise when an induction algorithm must revise a decision tree very often and oscillations between similar concepts decrease learning speed. We introduce a heuristic and an algorithm with theoretical and experimental backing to tackle this problem.