Technical documentation, such as user manuals and technical specifications, often constitutes the first line of support when users need help or want to learn more about a product or a service. It becomes an important part of user experience and the information provided needs to be of appropriate quality. Machine learning gets more and more important in all areas of software technologies. Applications range from system optimization and testing to adaptive software architectures. This project develops a machine earning approach to information quality assessment.
In our work we use machine learning to define what users consider high information quality. Therefore, a number of experts will judge quality of sample documents. At the same time, we automatically assess some attributes of these documents. This training data is input to classification (or linear regression) of algorithms. As a result, we get a classifier (or scorer) able to judge quality of yet unknown documents based on automatically assessed attributes.
The project is evaluating a novel hypothesis. It has both high scientific potential and practical relevance. It gives the insight knowledge in machine learning and artificial intelligence, one of the most exciting technologies that computer science has ever developed. Expected timeframes are Spring – Autumn 2014.
Main requirements:
· familiarity with machine learning
· scientific precision in working and documenting the results is a must
· familiarity with software metrics would be a plus
Your applications please send to:
Welf Löwe, Welf.Lowe@lnu.se