Reading some papers and articles about PageRank and HITS algorithm, I've figured out that there's a problem called topic drift problem. Googling it, (since I wanted to know more about it) I only found a few articles which has mentioned it but not in detail. I'd like to know is the problem caused by the fact that these algorithms ignore web page content and only conider inlinks (PageRank) and outlinks. But how does it happen? what are researches done by far to avoid this problem?
1 Answer
this is a pretty good discussion in this ref p123: Google's PageRank and Beyond: The Science of Search Engine Rankings Amy N. Langville & Carl D. Meyer, Princeton University Press.
A final disadvantage of HITS is the problem of topic drift. In building the neighborhood graph $N$ for a query it is possible that a very authoritative yet off-topic page be linked to a page containing the query terms. This very authoritative page can carry so much weight that it and its neighboring documents dominate the relevant ranked list returned to the user, skewing the results toward off topic documents. Henzinger and Bharat suggest a solution to the problem of topic drift, weighting the authority and hub scores of the nodes in $N$ by a measure of the relevancy of a query [26].
ref [26] is apparently:
Improved Algorithms for Topic Distillation in a Hyperlinked Environment Bharat, Henzinger, 21st ACM SIGIR Conference on Research and Development in Information Retrieval 1998
on p.2 they have another description/example of topic drift:
We often find that the neighborhood graph contains documents not relevant to the query topic. If these nodes are well connected, the topic drift problem arises: the most highly ranked authorities and hubs tend not to be about the original topic. For example, when running the algorithm on the query
jaguar and car
the computation drifted to the general topiccar
and returned the home pages of different car manufacturers as top authorities, and lists of car manufacturers as the best hubs.