Year 1 Ph.D. Plan of Study for Morten Goodwin Olsen

General Information

Abstract

The plan is prepared in September 2009 as a formal Ph.D. delivery for Morten Goodwin Olsen after one year study at the University of Aalborg.

It is often challenging to know which government web sites are working well and which have significant potential for improvements. There are very few unified ways of measuring such services, and these are often based on manual evaluations which are both costly and bias prone.

The expected outcome from this Ph.D. can be divided into two main categories:
(1) A specification and implementation of a tool which automatically benchmarks web sites according to accessibility, transparency, efficiency and impact. Several of the tests will utilize so-called learning algorithms $-$ algorithms which, based on some input or training data, will be able to perform tests which typically are done manually.
(2) Empirical results of accessibility, transparency, efficiency and impact, retrieved automatically using the implementation.

The significance will be the possibility to easily benchmark local government web sites. This will be an important tool for both raising and maintaining awareness on important topics such as accessibility, transparency efficiency and impact as well as identify good practices.

Accessibility:Web sites are often created in such a way that they limit some disabled users, such as blind people, from using the web sites. Measuring web accessibility means detecting barriers on web pages.

Transparency:A government agency can to a larger and smaller degree be transparent and open to the public. As an example, a transparent municipality would publish contact information, videos of political meetings, etc.

Efficiency:eGovernment efficiency can be seen as government value for money: how good eGovernment services can you get from the available funding.

Impact:Impact is often defined as the measurable effects for the citizens from an eGovernment service. Such an effect could be the percentage of a population which hands in tax surveys online.

For measuring online eGovernment services, web-mining techniques will be used, including so-called machine learning algorithms. The main idea is for these algorithms to automatically learn what is transparent, efficient, accessible and has an impact based on some input data, such as manual surveys. By doing this, the algorithms benefit directly from the manual data.

The key methods in this research are literature review, experimental prototyping and evaluation of web sites.