Piotr Andruszkiewicz

Probability Distribution Reconstruction for Nominal Attributes in Privacy Preserving Classification

Concerns about privacy of data used in data mining have emerged recently. Users are afraid of misuse of this data and discovered knowledge. Thus several methods of preserving privacy classification have been proposed in literature. One of these methods enables miners to use continuous and nominal attributes simultaneously in classification. Reconstruction of probability distribution is an important task in privacy preserving classification for both nominal and continuous attributes which were distorted with the randomization-based technique and are stored in centralized database. We present the new algorithm - EQ - for reconstruction of probability distribution of nominal attributes, which outperforms former algorithm especially for high privacy levels. Effectiveness of the new solution (information loss in reconstruction of probability distribution of nominal attributes and accuracy of classification) has been tested and presented in this paper.

Intelligent information processing for building university knowledge base

There are many ready-to-use software solutions for building institutional scientific information platforms, most of which have functionality well suited to repository needs. However, there have already been discussions about various problems with institutional digital libraries. As a remedy, an approach that is researcher-centric (rather than document-centric) has been proposed recently in some systems. This paper is devoted to research aimed at tools for building knowledge bases for university research. We focus on the AI methods that have been elaborated and applied practically within our platform for building such knowledge bases. In particular we present a novel approach to data acquisition and the semantic enrichment of the acquired data. In addition, we present the algorithms applied in the real life system for experts profiling and retrieval.

Lazy Approach to Privacy Preserving Classification with Emerging Patterns

In privacy preserving classification when data is stored in a centralized database and distorted using a randomization-based technique emerging patterns can be used to contrast classes. We present the new approach to privacy preserving classification for centralized data based on Emerging Patterns. In contrast to previous works, we use a lazy approach based on DeEPs to classification. Effectiveness of this solution has been tested and presented in this paper.

Privacy Preserving Classification with Emerging Patterns

In privacy preserving classification, when data is stored in a centralized database and distorted using a randomization-based technique, we have information loss and reduced accuracy of classification. This paper presents a new approach to privacy preserving classification for centralized data based on Emerging Patterns. The presented solution gives higher accuracy of classification than a decision tree proposed in the literature, especially for high privacy. Effectiveness of this solution has been tested on real data sets and presented in this paper.

AI Platform for Building University Research Knowledge Base

This paper is devoted to the 3-years research performed at Warsaw University of Technology, aimed at building of an advanced software for university research knowledge base. As a result, a text mining platform has been built, enabling research in the areas of text mining and semantic information retrieval. In the paper some of the implemented methods are tested from the point of view of their applicability in a real life system.

Home Pages Identification and Information Extraction in Researcher Profiling

In order to create a structured database describing researchers, home pages can be used as an information source. As the first step of this task, home pages are searched and identified with the usage of the classifier. Then, the information extraction process is performed to enrich researchers profiles, e.g., extract phone and e-mail. We proposed the algorithm for extracting phone numbers, fax numbers and e-mails based on generalised sequential patterns. Extracted information is stored in the structured database and can be searched by users.

Agent-Based Web Resource Acquisition System for Scientific Knowledge Base

The paper presents the summary of design, development, and deployment of the Web Resource Acquisition System as a mean to gather knowledge and scientific resources for common University Knowledge Base. This module was designed and developed under the SYNAT research project. The module uses common logical data interface developed for this purpose and is integrated with the user presentation layer of the Knowledge Base from the Warsaw University of Technology. The work emphasizes on the usage of definition and strategies in the context of Knowledge Delivery problem. Presented solution can be interpreted as an alternative to web crawlers when it comes to general problem of browsing through the Internet data. In particular, the effort was put on in-depth coverage of requested domain of knowledge when specifying query. At the same time, integration with the semi-automatic classification module was performed to support assessment of the retrieved resources with respect of their types. That resulted in development of Multi Agent System for universal resource delivery. Heterogeneous knowledge sources as Bing, Google, CiteSeer, etc. were used to provide wide-ranging input data from the Internet.

Reduction Relaxation in Privacy Preserving Association Rules Mining

In Privacy Preserving Association Rules Mining, when frequent sets are discovered, the relaxation can be used to decrease the false negative error component and, in consequence, to decrease the number of true frequent itemsets that are missed. We introduce the new type of relaxation - the reduction relaxation that enable a miner to decrease and control the false negative error for different lengths of frequent itemsets.

Automatic Extraction of Profiles from Web Pages

Web pages are usually unstructured and Information Extraction from them is not trivial. In the paper we describe the process of Information Extraction on the example of researchers’ home pages. For this reason we applied SVM, CRF, and MLN models. Performed analysis concerns texts in English language only.

How Rich Are Mobile Rich Internet Applications

Nowadays, mobile devices become more and more powerful and they offer continuously growing capability in terms of computing capability, size of screen, available memory, etc. It causes that delivering applications on mobile devices is even more attractive and draws attention of many software producers. However, variety of mobile devices and incompatibility of their operating systems makes very difficult and costly to implement an application with rich functionality which may be installed and used on different types of mobile devices. The solution to this problem can be Rich Internet Applications (RIAs) which are accessible via Internet browsers enhanced by the standardized functionality supporting RIAs.