REPOZITORIJ > REZULTATI

Doktorska disertacija

Zmanjšanje vpliva pristranosti uporabnikov v računalniških sistemih zaupanja

Avtor(ji): Tanja Pavleska (Avtor), Borka Jerman-Blažič (Mentor)

Datum zagovora: 15.06.2016

Organizacija: MPŠ - Mednarodna podiplomska šola Jožefa Stefana

PID: 20.500.12556/ReVIS-15045

Ogledi: 4 | Prenosi: 0

Povzetek

The functionality of the Internet has outgrown its initial goal to simply transfer information between distant sites. Influenced by phenomena from the traditional world, it largely cherishes the human desire to be part of a worldwide network. It is clear that networking among people dates from long before the commercial expansion of the Internet, but there is a new perception of socializing and business making created by the impression management of people and services on the Web. The advent of social networks opens up myriad opportunities for merging the social character of trust with the technical possibilities offered by information and communication technologies. The growth of user-generated content requires social approaches for information filtering and aggregation.
The success of trust-based methods depends on the research aimed at identifying when and in what manner these methods offer most benefit. This work is a contribution to that purpose. We believe that such analysis may result in online networks providing reliable means for building trust and reputation on the Web. Moreover, the thesis aims to testify that trust awareness is needed on the Internet as it is in the real world.
This work determines the systemic features of trust and introduces a novel framework of design properties based on the principles of systems theory. We regard trust systems as human-centric, not merely technical (computational), and address some of the organismic properties that current schemes for modelling trust systems disregard. For that purpose, we explore and exploit the role of transitivity, subjectivity and user bias in trust systems and investigate their implications for the design of the computational systems that cherish the digital analogue of social trust. Through case studies, semantic, empirical and statistical analysis we identify several issues with the existing approaches that tackle trust-related problems, whereas through models based on subjective logic, systems science and balance theory we propose ways to address these problems in a multidisciplinary manner.
More concretely, this thesis demonstrates the significant discrepancy between the aim of implementation of the online trust systems and the users’ perception of those systems. By employing systemic analysis, we identify four additional factors that contribute for the completeness of a novel framework for trust systems design. Then, by performing empirical study on an existing system, we demonstrate the presence of several concrete user biases that stem from, but also contribute to the discrepancy between the trust system implementation and the users’ perception and behaviour. Understanding this relation by the system designers is important for reducing the complexity and improving the user-experience and the system-performance. The results and the analysis are then used to propose a methodology for user bias identification and mitigation in the form of a choice architecture for trust systems and a formal model based on subjective logic. Several extensions to the model are also proposed and placed in the plan for future work.

Priloge

Citiraj to delo