ir.webutils
Class LinkHeuristic

java.lang.Object
  ir.webutils.LinkHeuristic

public class LinkHeuristic
extends java.lang.Object

Evaluates a web link (ScoredAnchoredLink) based on satisfying a set of "want strings" and "help strings".

The existing search heuristic considers four factors in order of importance:

1. (WC) The number of the want-strings that are found
2. (WT) The total number of times a want-string is found
3. (HC) The number of the help-strings that are found
4. (HT) The total number of times a help-string is found

A page is scored as

S = 1000*WC + 100*HC + 10*WT + HT

A link is scored partly based on the text appearing directly in the link and partly based on the surrounding page. If L is the S score for the text in the link and P is the S score for the overall page, then a link is scored as

L/2 + P/2

getting half its score from it's own text and half from its surrounding page.

Field Summary

java.lang.String[]	helpStrings The array of help strings to help find the want strings
java.lang.String[]	wantStrings The array of want strings that are desired

Constructor Summary

LinkHeuristic()
Construct an empty heuristic

LinkHeuristic(java.lang.String[] wantStrings, java.lang.String[] helpStrings)
Construct a heuristic with the given wantStrings and helpStrings

Method Summary

double

scoreLink(ScoredAnchoredLink link, HTMLPage page) Heuristically score the given link appearing on the given page

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

wantStrings

public java.lang.String[] wantStrings

The array of want strings that are desired

helpStrings

public java.lang.String[] helpStrings

The array of help strings to help find the want strings

Constructor Detail

LinkHeuristic

public LinkHeuristic()

Construct an empty heuristic

LinkHeuristic

public LinkHeuristic(java.lang.String[] wantStrings,
                     java.lang.String[] helpStrings)

Construct a heuristic with the given wantStrings and helpStrings

Method Detail

scoreLink

public double scoreLink(ScoredAnchoredLink link,
                        HTMLPage page)

Heuristically score the given link appearing on the given page