Goodbye PR, Hello TSPR

[glengara]

G is in the process of dropping Page(the man) Rank in favour of Topic Sensitive Page(the thing) Rank, a quantum leap in search relevance and something many of us have been urging G should do.

The object of both Hilltop and Localrank is to determine TSPR, Hilltop through off-page factors, Localrank through on-page factors, and an amalgamation/combination of both, would work best.

The starting step for both is a search query, and CIRCA, the Applied Semantics methodology will be used to return the initial "corpus" of relevant documents both need to determine TSPR.

The mystery remains why we have all been invited to watch this slow, painful, and at times ludicrous transformation.

All strictly IMO, of course ;-)

[SVB]

You from Dunleiorgy in ireland?

[glengara]

Yep ;-)

[SVB]

woohoo :-D
Knew you were Irish from your name and the "Done Leery" :-D

So, what does this TSPR mean to webmasters. How do we use it etc?

[Rob]

Would be good to see on-page factors get more weight.

Still, a large enough amount of inbound links can get you on top of any search term. They just seem to favour links from related pages.

[glengara]

* How do we use it etc? *

Good question, to which I have no quick answer ;-)
Relevance or topicality appears to be the key with both, particularly with off-page factors.
It seems we'd need links from "expert" pages ( Hilltop) as well as links from similar content sites (Localrank), and topical content would appear to be the major factor in both.

Internal and off-topic external links would be discounted, sub-domains and multi-sites on the same topic harder to manage.

[compar]

Hilltop through off-page factors, Localrank through on-page factors

Can you illustrate how you came to this understanding? I think both these techniques are doing the same thing. They have both been authored in part or in total by the same person -- Krishna Bharat. The wording is slightly different, otherwise Google would have been directly applying for a patent on Hilltop, but what Hilltop calls "expert documents", Local Rank calls a "corpus".

The main difference I see between these two methodologies is that Hilltop seems to specifically suggest that directories will play a large part in it's body of "expert documents", where as Local Rank never gives any detail on how their "corpus" is to be determined.

But to my reading once the search is initiated both systems do the same thing. They look at how many links from their preselected set of documents go to the various pages in the intitial ranking, eliminate any affiliate links in that subset, and assign relevance values based on the linking relationship to the "expert documents" or "corpus".

[glengara]

Hi Bob, thought I'd see you here ;-)

*But to my reading once the search is initiated both systems do the same thing. They look at how many links from their preselected set of documents go to the various pages in the intitial ranking, eliminate any affiliate links in that subset, and assign relevance values based on the linking relationship to the "expert documents" or "corpus".*

Absolutely concur.

IMO, the difference between Hilltop and Local rank is in how each determines their respective "expert" pages.
Hilltop: initial search query brings up relevant pages with on-topic out-going links.
LocalRank: initial search query brings up highly relevant pages, based on on-page factors.
see:
Detailed description - (Fig 1)
"For example, documents may have their rank value based on the proximity of the search terms in the document (documents with the search terms close together are given higher rank values) or on the number of occurrences of the search term (e.g., a document that repeatedly uses a search term is given a higher rank value)."

As I mentioned in another thread, I don't see either one giving us a "quantum leap in search relevance", but together, they very well might.

[compar]

Detailed description - (Fig 1)
"For example, documents may have their rank value based on the proximity of the search terms in the document (documents with the search terms close together are given higher rank values) or on the number of occurrences of the search term (e.g., a document that repeatedly uses a search term is given a higher rank value)."

For some reason I'm having trouble seeing Fig 1 at the moment. But I have this question. What documents are they talking about at this point? Are they talking about the documents in the "corpus" the comparison or "expert" set. Or are they talking about the on page or content factors on the pages that they are considering for ranking in the SERP?

If they are talking the "corpus" then there is no different between what the two systems are attempting to do. If they are talking about the pages in the eventual SERP then there is a difference.

If they are talking about this latter situation it puts content squarely into the SEO formula.

BTW I still think if Google is using any of this it is Local Rank. That what they hold the patent for. Could they use Hilltop without paying royalties and/or being in jeopardy of a law suit?

[PhilC]

Sorry, but I think there's a misunderstanding about how both Hilltop and LocalRank work.

A Hilltop system gets its results set from a database of expert pages. The pages that the expert pages link to are the ones that form the results.

LocalRank simply rearranges a results set, however it is derived, according to any linkages between the pages in the set. LocalRank doesn't generate a results set.

<added>

Originally Posted by glengara

The object of both Hilltop and Localrank is to determine TSPR, Hilltop through off-page factors, Localrank through on-page factors

Hilltop determines things by on-page factors. LocalRank determines things by off-page factors - links to other site's pages.
</added>

[glengara]

*Are they talking about the documents in the "corpus" the comparison or "expert" set. Or are they talking about the on page or content factors on the pages that they are considering for ranking in the SERP?*

IMO they're talking about the pages that will form the basis of their "experts" for LocalRank.

*Could they use Hilltop without paying royalties and/or being in jeopardy of a law suit?*

Doesn't he work for G? He might even get a kick out of them using it.....

[glengara]

* LocalRank doesn't generate a results set.*

"Finally, the method refines the relevance scores for the documents in the generated set based on the local score values."

Reads to me like it does, Phil.

[glengara]

<added>
*LocalRank simply rearranges a results set, however it is derived*

The way I see it, how it is derived is the crucial matter here.......

[PhilC]

Originally Posted by glengara

* LocalRank doesn't generate a results set.*

"Finally, the method refines the relevance scores for the documents in the generated set based on the local score values."

Reads to me like it does, Phil.

LocalRank takes a pre-compiled results set and uses the linkages between the pages in the results set to arrives at local scores.

[glengara]

"13. means for refining the relevance scores for the documents in the initial set based on the local score values."

So what about * LocalRank doesn't generate a results set.* ?

[PhilC]

I'll leave you to it. LocalRank works on a results set; it doesn't compile one itself. I'll leave you to it.

[glengara]

Both Hilltop and Localrank work on a results set.....

[kkibak]

Hi guys,

I realize I'm jumping into an expert debate/conversation here, but I need something cleared up.

Am I correct in understanding that LocalRank basically re-sorts a list of results that google, or whatever engine, finds originally?

In other words does this mean that if I was targeting the term "knives," Google would compile a result list for "knives" and then compile a second set of results based on the links in the first list?

I'm sorry if I'm totally off in this, but just trying to get the basics of this LocalRank stuff cleared up as it's all way over my head at the moment.

[compar]

In my understanding both Local Rank and Hilltop do a secondary ranking based on the relationship of the pages in the first ranking to a preselected set of "expert documents" as Hilltop calls them or a "corpus" of pages as Local Rank calls them.

What they absolutely have in common is the removal of pages called "affiliate" pages from the preselected set.

This makes a lot of sense to me and explains why immediately after Florida you could get the original ranking by using a "-garbageterm" with your search. Apparently the "-gargabgeterm" negated or turned off the second ranking.

[glengara]

Not sure where the idea of "pre-selected" or "pre-compiled" comes from, found nothing about it in Hilltop, and LocalRank only mentions a predefined number.

"5. The method of claim 3, wherein a predefined number of the documents in the sub-set are used to calculate the local score value."