Thanx Ken :)P -----Original Message----- From: Ken Krugler [mailto:[EMAIL PROTECTED] Sent: Monday, September 05, 2005 8:10 PM To: [email protected] Subject: RE: Link Analysis Score..
>Thanx for the information....I went through Distributed Analysis Tool--- > >What is this next score? What is its significance? That's a good question for Mike Cafarella. My simplistic understanding is that the basic algorithm is using score distribution (each page's score is divided among all of the pages it directly links to). nextScore propagate scores to pages that in turn contain outlinks, but that's the part that I don't understand. >One more thing... >Once I get the Link Analysis Score, I set it as a boost value in the >boost field. >Is it so? If you run the DistributedAnalysisTool, that alter the page scores in the WebDB. Then if you run the UpdateSegmentsFromDB tool, that will update the fetcher output subdirectory inside of a segment directory, using the revised page scores. Finally, if you then generate the Lucene index, these updated page scores will be used when calculating the Lucene document boost. Or at least that's how I think it works :) -- Ken >-----Original Message----- >From: Piotr Kosiorowski [mailto:[EMAIL PROTECTED] >Sent: Monday, September 05, 2005 1:33 AM >To: [email protected] >Subject: Re: Link Analysis Score.. > >There are many ways nutch can boost document in the index. But I suspect > >you are refereing to analyze process - it uses PagrReank computation for > >page score. For details read DistributedAnalysisTool - especially >computeRound method. >Regards >Piotr >Rozina Sorathia wrote: >> I wanted to know where exactly the Link Analysis Score is calculated >...Is >> there any code snippet available.? >> >> How is the Link Analysis Score affecting the overall final score of >the >> document? >> >> >> >> >> >> >> >> //Rozina Sorathia,// >> >> //Systems Executive,// >> >> //KPIT Cummins Infosystems Ltd.,// >> >> //[EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>// >> >> >> >> >> -- Ken Krugler TransPac Software, Inc. <http://www.transpac.com> +1 530-470-9200
