Julia Hoxha

Open Government Data on the Web: A Semantic Approach

Initiatives of making government data open are continuously gaining interest recently. While this presents immense benefits for increasing transparency, the problem is that the data are frequently offered in heterogeneous formats, missing clear semantics that clarify what the data describe. The data are displayed in ways, which are not always clearly understandable to a broad range of user communities that need to make informed decisions. We address these problems and propose an overall approach, in which raw statistical data independently gathered from the different government institutions are formally and semantically represented, based on an ontology that we present in this paper. We further introduce the approach deployed in publishing these data in alignment with Linked Data principles, as well as present the methods implemented to query single or combined dataset and visualize the results in understandable ways. The introduced approach enables data integration, leading to vast opportunities for information exchange, analysis on combined datasets, simplicity to create mashups, and exploration of innovative ways to use these data creatively.

open.data.al: increasing the utilization of government data in Albania

Open Data practice requires that data are freely available for everyone, laying the foundations for transparency and decision making. Many democratic countries have supported this practice, freeing and facilitating access to their government data. Open Data Albania is a new initiative that embraces the principles of Open Government Data for Albania. This project aims to collect from public offices data on socio-economic indicators, process and publish them based on semantic technologies and Linked Data principles. This work presents the approach we have deployed in publishing governmental data as linked data and the methods implemented to query datasets and visualize the results. In our approach, we aimed for quality of the published open data, designing cases of problems with great interest for the community. Each case is offered in our website as an article accompanied not only with the statistical dataset, but also with graphic visualization and a detailed textual description.

Semi-automatic Acquisition of Semantic Descriptions of Processes in the Web

Most of today’s business processes are complex and consist of more than one party or single step procedures. In the Web, this is reflected by the existence of billions of Web sites, which may be regarded as complex processes, and on the other side only a few thousands of publicly available WSDL files that present single services. The availability of semantic descriptions of services and processes in the Web facilitates their discovery, as well as their composition into more complex workflows. It also facilitates the automatic execution of such workflows despite their heterogeneity. However, the deficit of semantic descriptions of Web processes deprives the users from using such sophisticated automatic techniques. The scope of our research is to fill this gap by providing semiautomatic techniques for the acquisition of a large number of semantic process descriptions on the (deep) Web. We model the data found in the online sources using ontologies, mine the process a user follows through the Web forms and generate a semantic description of this process. We present in this paper the implementation of our algorithms for the acquisition of process descriptions. We also provide a Web-based editor for manual annotation of new processes and refinement of the automatically-generated descriptions.

First-Order Probabilistic Model for Hybrid Recommendations

In this paper, we address the task of inferring user preference relationships about various objects in order to generate relevant recommendations. The majority of the traditional approaches to the problem assume a flat representation of the data, and focus on a single dyadic relationship between the objects. We present a richer theoretical model for making recommendations that allows us to reason about many different relations at the same time. The model is based on Markov logic, which is a simple and powerful language that combines first-order logic and probabilistic graphical models. We apply a hybrid, content-collaborative merging scheme through feature combination. We experimentally verify the efficacy of our theoretical model, and show that our method outperforms state-of-the-art recommendation approaches.

Learning Relevance of Web Resources across Domains to Make Recommendations

Most traditional recommender systems focus on the objective of improving the accuracy of recommendations in a single domain. However, preferences of users may extend over multiple domains, especially in the Web where users often have browsing preferences that span across different sites, while being unaware of relevant resources on other sites. This work tackles the problem of recommending resources from various domains by exploiting the semantic content of these resources in combination with patterns of user browsing behavior. We overcome the lack of overlaps between domains by deriving connections based on the explored semantic content of Web resources. We present an approach that applies Support Vector Machines for learning the relevance of resources and predicting which ones are the most relevant to recommend to a user, given that the user is currently viewing a certain page. In real-world datasets of semantically-enriched logs of user browsing behavior at multiple Web sites, we study the impact of structure in generating accurate recommendations and conduct experiments that demonstrate the effectiveness of our approach.

Semantic Formalization of Cross-Site User Browsing Behavior

Large amounts of data are being produced daily as detailed records of Web usage behavior, but the task of deriving actionable knowledge from them remains a challenge. Investigations of user browsing behavior at multiple websites, while more beneficial than studies restricted to a single site, still need to tackle the problems of information heterogeneity and mapping usage logs to meaningful events from the application domain. Focusing on the problem of modeling cross-site browsing behavior, we present a formalization approach based on a Web browsing Activity Model (WAM). We introduce a novel two-staged approach for the semantic enrichment of usage logs with domain knowledge, bringing together Semantic Web technologies and Machine Learning techniques. For learning the semantic types of logs, we present a supervised multi-class classification formulation, deploying structural Support Vector Machines with new sequential input features. We provide an implementation of these approaches and show the results of evaluation with real-world data.

Defining Digital Library

This paper reflects on the range of the definitions of digital libraries demonstrating their extent. We analyze a number of definitions through a simplified intensional definition method, through which we exploit the nature of the definitions by analyzing their respective genera and attributes. The goal of this paper is to provide a synthesis of the works related to definitions of digital library, giving a fine-grained comparative approach on these definitions. We conclude that, although there are a large number of definitions, they are defined in overlapping families and attributes, and an inclusive definition is possible.