Links tell us about lexical and semantic Web content

TL;DR

This paper formalizes and validates two conjectures linking Web link structure to content, showing that links can reveal lexical and semantic information, which explains the effectiveness of modern search technologies.

Contribution

It introduces and empirically tests the clink-content and link-cluster conjectures, connecting Web topology to content inference methods.

Findings

Pages are similar to linking pages in lexical content

Web communities show heterogeneous link-content correlations

Pages about the same topic tend to cluster together

Abstract

The latest generation of Web search tools is beginning to exploit hypertext link information to improve ranking\cite{Brin98,Kleinberg98} and crawling\cite{Menczer00,Ben-Shaul99etal,Chakrabarti99} algorithms. The hidden assumption behind such approaches, a correlation between the graph structure of the Web and its content, has not been tested explicitly despite increasing research on Web topology\cite{Lawrence98,Albert99,Adamic99,Butler00}. Here I formalize and quantitatively validate two conjectures drawing connections from link information to lexical and semantic Web content. The clink-content conjecture states that a page is similar to the pages that link to it, i.e., one can infer the lexical content of a page by looking at the pages that link to it. I also show that lexical inferences based on link cues are quite heterogeneous across Web communities. The link-cluster conjecture states that pages about the same topic are clustered together, i.e., one can infer the meaning of a page by looking at its neighbours. These results explain the success of the newest search technologies and open the way for more dynamic and scalable methods to locate information in a topic or user driven way.