Forum Moderators: open
Authority is built from inbound and outbound relevant links. Add a dash of pagerank, relevant link text and voila, there's a measure of a site's authority :)
There's no quantitative method available to us to measure relevance and authority, so you kind of have to play it by ear. But if you look at the sites ranking well in your major SERPs, then you can get an idea of what the most authoritative/relevant sites for those topics are. Just need to set up some linking with those places and you'll be full up on authority/relevance :)
For example, a site about polyester fiber might be relevantly linked to by anyone in the textile industry (clothiers, yarns, weaving, etc.), anyone in the craft industries, anyone in the agricultural industries (bailing twine), the petro-chemicalindustries, and sites from the sporting goods industry (flytying, etc.).
Pretty diverse, and really only the beginning.
The same is true of many other types of sites. A site selling swords and cutlery might recieve links from kitchen-ware sites to fantasy role playing sites to movie fan club sites to sporting goods, etc. All potentially relevant.
If too much weight is put on algo derived relevancy then I would expect that there would be some deterioration in the SERPs.
WBF
I doubt that. Your examples suggest the opposite. They suggest an authoratative domain will rank for several things, for good reason. A sports website might get linked from baseball, fencing and curling sites. That wouldn't deteriorate anything. It would on the contrary be evidence that the domain in question is far more likely to have good content, because it is deemed trustable on more than just one limited topic.
This is the core of good search results. The key element though is in the search engine recognizing *some* commonality between curling and football. Then this sports niche authority with varied links should get the nod over CNN, which would also have varied links but most would look out of place if listed as "similar pages".
It's a question of the engines getting much better at stemming, related words, similar pages and recognizing genuine niche/theme.
It's a question of the engines getting much better at stemming, related words, similar pages and recognizing genuine niche/theme.
This would, IMHO, require a level of AI that we are far, far from achieving.
We are really talking about free association, which as far as I know remains in the sole domain of human consciousness. It would probably need some form of fuzzy logic, and a mirroring of the human experience of the world. Simple semantic analysis is as unlikely to accomplish this as the structural analysis that was used on page by the early SE's and that was expanded to the web as a whole when G started using link analysis.
You or I could develop quite a list of relationships to the word "cork". Baseball, wine, tree, bark, trivet, coaster, bumber, float, tackle, boating, stopper, boots, oak, bulletin board, Portugal, Santa Rosa Junior College, etc. Where is the information technology that can do this and apply it across billions of web pages?
When/if it is developed, I would agree that there would be an improvement in SERPs, but I doubt that I will live to see it.
WBF
So these links are 100% relevant but don't really mean anything (or at least shouldn't).
I guess things are right back to straight forward PR in this kind of industry with a few exceptions such as bona-fide travel/hotel sites.
This will help you to know which terms will be found 'relevant' to the keywords you wish to target.