Back to search

University of Patras

Μοντελοποίηση του κοινωνικού πλαισίου για τη διαχείριση προσωπικής πληροφορίας σε συσκευές κινητού υπολογισμού

Abstract

dc:description

In an ideal application of personal information management, users should always have available the appropriate information to the right place and in the right format, at that quality and completeness that is necessary to meet their needs. This requirement becomes even more urgent in the case of users with mobile computing devices, which face limitations compared to desktop computers that increase the challenges and complicate personal information management. Despite the technological advancements regarding computing power, storage capacity and connectivity, the screens of mobile computing devices remain significantly smaller than those of desktop computers, reducing the amount of information that can be presented simultaneously, while reading a large amount of information on these devices requires frequent horizontal scrolling and focused concentration. Moreover, the equipment and applications that are integrated in these devices allow the easy creation of big volumes of personal information data, such as videos, photos, notes, etc. Also, the context of use of these devices is often changing (e.g. while users are on the move their location changes, as well as the devices and people with whom they interact, their objectives and their needs) and it is more demanding than in the case of desktop use. The above characteristics, combined with the reluctance of users to devote time and effort to delete personal data result in information retrieval difficulties. On the other hand, mobile computing devices are equipped with a multitude of sensors that record the context of the user (mobile context): location, activity (e.g. walking, jogging, on a call), noise levels, environmental conditions etc. Furthermore, as users consider them as trusted devices and often leave them always on, they contain large amounts of personal information related to the users’ social environment. This information is often generated automatically. For example, for each call made or received on a mobile phone a relevant record is saved in the call list, which includes the date and time, the contact and the duration. Also, part of this information is generated by the users themselves: SMS and MMS, photos, videos and audio files, calendar events, contacts etc. Therefore, these devices except from the users’ mobile context are also informed about their social environment (social context). The combination of mobile and social environment results in a dynamically defined context, known as mobile social context. Personal information is highly related to the context of use, which is also true for its management activities. As the information needs of the users depend not only on mobile context but also on their social interactions, the aim of this PhD research was to create a model that uses the contextual information collected by mobile computing devices in order to improve the management of personal information. The stage of personal information management lifecycle on which this research had focused was exploitation. The case of mobile contacts was studied as a representative problem domain, because their number is constantly increasing, while users are reluctant to devote time and effort to organize or delete them and as the main mission of these devices in their most common form (Smartphones) is to facilitate communication. The goal of the research was to investigate in which way context can improve contact retrieval when starting a communication session, especially for the case of outgoing calls. However, the presented approach can be extended and applied to any other personal information element (e.g. messages, emails, notes, photos, video, etc.). The structure of this thesis is as follows: Initially, Chapter 1 introduces the reader to the problem of personal information management and analyzes the reasons why the studied problem shows an increased interest for the case of mobile computing devices. Next to linking the problem with mobile social context, the framework of the PhD research is presented: motivation, objectives and research questions. Chapter 2 presents an extensive literature review of related research activities for personal information management in mobile computing devices and use of context in general, and particularly for the case of mobile contacts. Chapter 3 proposes a model of context representation in the form of vector. Although this is a relatively simple representation, it allows for the easy context management and enables future integration of new dimensions. Moreover, it allows for the application of dimensionality reduction algorithms. A concept that was investigated was the reduction of the context dimensions to 2 or 3, a practice which would allow representation of personal information objects in two-dimensional or three-dimensional space and could possibly improve retrieval. Eventually, the idea was abandoned, as the number of adopted dimensions were not such as to justify the need for application of such techniques. In order to identify the context dimensions that are relevant to the problem of contact retrieval when starting a new call session, an experiment that was conducted and included analysis of real data from users’ devices is presented. The analysis identified two dimensions as important (frequency and recency of communication) and rejected others that are often used in literature, nevertheless without sufficient justification. Moreover, it revealed interesting communication patterns of the users, such as a form of Pareto Principle regarding the number of calls to contacts. Chapter 4 presents the algorithm that was developed for the retrieval, based on the ranking and selection of contacts using frequency and recency of communication. The algorithm, which operates as a prediction algorithm of the N contacts that are most likely for the user to call, was assessed by simulation on the available datasets. The results are more or equally successful than those that are reported in relevant research papers. In addition, some of the available algorithms in literature were implemented and were directly compared to the proposed algorithm, using the same dataset. This comparison resulted in much better performance of the proposed solution. Furthermore, three different types of users were identified according to their perceived “socialness”. For low or medium “social” users the performance of the algorithm is extremely high, while for more “social” users there is still room for improvement. In addition, in this chapter the results of a field experiment that was conducted to evaluate the algorithm under real conditions are presented, confirming the conclusions of the simulation procedure. Chapter 5 presents a study on the impact of location on contact retrieval. Location is considered as a very important context dimension and is very often suggested as a candidate dimension for call prediction in literature. Initially, physical location was analyzed, without finding a significant improvement in the prediction results when incorporating this dimension. Then, semantic location was also considered in the prediction algorithm, leading to similar conclusions. Similar conclusions resulted from the integration to the algorithm of contextual dimensions that are indirectly related to location, such as temporal context (time of day). Although we must be cautious in generalizing these findings, it seems that location is not such a useful context factor for many users regarding contact retrieval, as perceived in the scientific community. Finally, this thesis concludes with an epilogue where the conclusions are summarized, the research questions are answered and the contribution of the research is emphasized. At the same time, a generalization of the proposed model that can be applied to any type of personal information items or even to a combination of different types is presented.

Degree

thesis:*
Grantor dc:publisher
University of Patras
Year dc:date
2016

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Plessas, Athnasios
  • Πλέσσας, Αθανάσιος

Subjects

dc:subject × 16

Rights

Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/50046

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Plessas, Athnasios; Πλέσσας, Αθανάσιος. Μοντελοποίηση του κοινωνικού πλαισίου για τη διαχείριση προσωπικής πληροφορίας σε συσκευές κινητού υπολογισμού. University of Patras, 2016. http://hdl.handle.net/10442/hedi/50046