Back
to indexMonika Henzinger
Summary
This article describes query-independant and query-dependant link analysis algorithms. Query-independant algorithms "measure the intrinsic quality" of a page based on the number of links to a page. The PageRank algorithm used by the Google search engine uses this technique and includes a recursive weighting of the quality of the page that linked to it. Needless to say, Google works very well with this technique. The second query-dependant algorithm uses the neighborhood of a page that is related to a topic to iterateively determine the most relevant page. This iterative technique identifies the authority and the hub scores for a particular document, and recomputes their scores by adding up their hub and authority scores, respectively. This technique is evaluated by others, and determined to be robust but intrinsically is subjected to topic drift and authors may intentionally be able to skew the results in their favor. This article is able to concretely present an overview of the link analysis algorithms, and provides a clear picture of the complications of their designs. It also provides a direction for researching the algorithms in more detail, and their critiques. However, there is little technical content from which future work can be built.
Methods
This paper provided a comparison of techniques, and did not provide any algorithms directly. The wise approach would be to research both the PageRank and HITS algorithm directly.
Keywords
link analysis, PageRank, HITS, query-dependant, query-independant, hub, authority, topic drift
Assumptions
Quality of pages is determined by the number of links to them Links to pages indicate a page that is related to the current page in some way
Rating
6
Bibtex Entry
@article = { henzinger00,
author = "Monika Henzinger",
title = "Link Analysis in Web Information Retrieval",
journal = "IEEE Data Engineering. Bulletin",
volume = "23",
number = "3",
pages = "3--8",
year = "2000",
url = "http://www.research.microsoft.com/research/db/debull/A00sept/henzinge.ps"
}