Back
to indexDell Zhang and Yisheng Dong
Summary
This paper presents a page ranking algorithm that leverages the user's browsing history, including time and links explored on particular pages, to rank the utility of a particular page. The foundation of this approach lies in the use of a Markov model to map the user's browsing history. All pages are sequentially visited, creating a Markov chain. Then, the probability of a user visiting a sequential page can be calculated.
The probabilistic nature of this algorithm, and the time required to generate the baseline data make it difficult to apply to the web search agent paradigm. In addition, this paper does not present any experimental evidence of their findings, or a direct application from which the prediction can be leveraged. Failure to incorporate the context of words or their lexical structure make it additionally difficult to apply in practice.
Keywords
web, resource discovery, search engine, rank algorithm, markov model, authority, hub, linear equations
Methods
The user's state as he browses through the web is a temporal sequence that is represented by a Markov chain. Links from a page can be appropriately ranked based on the probability, and the linkage structure. This theory assumes that what the user is currently browsing will predict what he will browse in the future. This idea is captured in the user's tendency matrix, where each matrix entry corresponds to a either the relevance, authority, integrativity (backlink count) and novelty real values. When this matrix is used in conjunction with the Markov chain, a transition probability matrix may be generated for each link, and the predicted next link will be the highest ranked result.
Assumption
What the user is currently browsing is indicative of what he will continue to browse.
Rating
5
Bibtex Entry
@article = { zhang00,
author = "Dell Zhang and Yisheng Dong",
title = "An efficient algorithm to rank web resources",
journal = "Computer Networks",
volume = "33",
pages = "449--455",
year = "2000",
url = "http://www9.org/w9cdrom/251/251.html"
}