www.MySearchEngine.here :)

[AlanMCSD]

You know, if you give 2 very good developers the task of creating a prototype for project A, you usually get back two very good solutions, and the longer they tweak the closer the 2 solutions become.

So I started thinking, to better understand search engines, I need to build one. Then and only then would I have a true appreciation for the nitty gretty details.

In a perfect world I would hire around 2 million people, to manually view, categorize, rate, and otherwise enter the information into a SQL DB running on a Data Center Cluster. 3 billion later I would be done, and have to start all over again. Any investors?

A more realistic approach would be done programmatically, and here are a few determinations that I have made for building MY search engine. This project actually seems like fun to me.

A) All tags including meta tags would be disregarded, here’s why. I don’t trust the masses to provide accurate information. I know that if I place relevance on a particular keyword, that keyword would be used excessively throughout a page. That would influence content in a negative way. A better approach is an algorithm to classify a site into general categories, sub categories, and so on based on the overall scheme of all the combined pages of the site. In short it would answer the question, what is this sites main Topic, area of specialty, etc?

B) A second algorithm would then define the relationship of external entities to this Topic. For instance “ping pong and billiards” would eventually cascade down a tree linking to games>>table top games. So at the end of a and b if some one searches for “pool tables” they should see table top games of the pool table variety. If someone searches on pool I would ask them to be more specific – yeah, I would actually prompt the user to provide more relevant information, instead of displaying 2,100,123 results and I don’t care.
I know there are tons of studies done on entity relationships – the main thing here is that I don’t want the site owner to tell me what the site is about, I need to figure that out programmatically.

C) The term page rank would be dismissed completely in favor of two terms (relevance and popularity)

D) Search relevance would be determined through the data accumulated in algorithm B.

E) Page Popularity would be determined by the number of inbound links, with the popularity of those backlinks taken into account.

F)Results would be ordered based on relevance, popularity.

G) G, Gee – I guess I have my work cut out for me.

In closing I would like to add that I know a lot you here, me included, have relatively new sites, and I am sure that many of you are providing better content and service then the top 10 search results list, and that to me is the most unfortunate thing about the whole internet. The best content, the best deals, and the best service providers are more often then not, over looked.