Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
With today’s rapidly growing Internet it is often difficult to find something one is looking for, even with the existing search engines. When one searches for information about a specific subject, the search results often contain thousands of choices of more or less related documents. Because of word ambiguities, some results can even be totally wrong. One could try to narrow the search with Boolean queries, but often users do not have the skills or willingness to do complex queries. A study by [Jansen] shows that an average query contains less than 3 keywords and that Boolean queries are rarely used. When users are faced with a large number of choices they like to turn to something that can help them to make a choice. Examples of this could be newspapers with recommendations of movies or restaurants. A good recommendation service for web pages would probably help many and save a lot of time and effort.
This report will describe and compare different methods and models that can be used to create a system that recommends web pages to users with the help of an automatically adapted personal profile. A prototype of such a system is also described and evaluated. The tests on the prototype show that the system is adapting to a user’s feedback and that the structure of web documents can be used to improve the effectiveness.
Place, publisher, year, edition, pages
2001. , 94 p.
IdentifiersURN: urn:nbn:se:kth:diva-93287OAI: oai:DiVA.org:kth-93287DiVA: diva2:515547
Subject / course
Master of Science in Engineering - Computer Science and Technology
2001-02-13, Biblioteket, IMIT, Isafjordsgatan 22, Kista, 11:00 (Swedish)
Maguire Jr., Gerald Q., professorHjelm, Fredrik
Maguire Jr., Gerald Q., professor