|News May 5, 2005
Relevancy. It's something that drives the search engine industry in almost every aspect. From providing quality search results to contextual advertising,
relevancy is crucial to any search engine's success. The question is how do you define relevancy? More importantly, how do search engines define relevancy?
How do you define relevance when it concerns search results? Which engine do you find to be most
relevant? How do you think relevance can be improved upon by the engine and the searchers alike? Discuss at WebProWorld.
These questions continue to give the debate life because what may be relevant to one person may not be relevant to another user and/or a search engine.
So what determines relevancy? Because this subject has such a gray area, there are no clear-cut answers. Search engines have their own methods for
determining relevancy and it's usually based on their respective algorithms. However, we've all dealt with queries containing results that leave you
scratching your head in confusion. Of course, there are other searches that can yield exactly what you are looking for. But why the disparity?
Obviously, all search results are predicated on the keyword(s) used in the query. If you are using general terms
business), you can expect convoluted or less targeted
search results. By the same token, if you are searching a more specific query
House of Flying Daggers DVD), odds
are the results will be more relevant.
When dealing with search engines and relevancy, I'm reminded of a quote featured in a
SearchEngineWatch blog entry by Greg Price. While discussing relevancy and search engines, Greg offered this: "Perhaps Udi Manber said it best at
PC Forum a few weeks ago when he told the audience that search engines are not mind readers." Due to the lack of clairvoyance on the part of the search
engine industry, Greg suggests learning how to properly refine search queries.
In light of the relevancy debate, Barry Schwartz of RustyBrick.com
had an idea concerning a test to determine what engine is the most relevant. In order to conduct this test, Barry developed
RustySearch.com, a "white labeled" search engine
"that randomly select(s) results from one of the top four engines and ask(s) you to
rate the search engine results, individually, from one to five."
RustySearch.com is not only going to help measure the relevancy of the big four (Google, Yahoo, MSN Search, and Ask.com), Barry is going to factor these
rankings and announce which search engine is the most relevant (does the winner receive a prize ;)?). In order to cut down on external biases, Barry also
made sure to hide which search engine supplied the results that are being ranked.
Currently, Barry has not revealed when the relevancy competition will be complete, so please stay tuned.
However, until the relevancy test is decided, other avenues to consider when it comes to search engine relevancy are the vertical search engines. Vertical
search engines are considered niche because they normally subject specific. For instance, the Thomas
Global Register is designed to search industrial-based products, while Jayde.com concentrates on the B2B
(business-to-business) world. If you were searching for something within these areas of interest, you would be better served using an engine that concentrates
on that subject.
However, just like Greg points out, very few of the average Joe Internet users are aware of such tools. That's why he stresses refining your search keywords.
Hopefully, this practice will return results that are more relevant to your area of interest.
Another factor that can impact the relevancy of
results has to do with the multiple meanings of individual words. This is pointed out quite well by Black_Knight on the
SEW forums, who posts: "When someone searches for 'thunderbird' there
are so many possible options. Do they mean the car, the TV puppet show, the drink, the email client, the mythical beast, or something else entirely? There
isn't enough contextual info in the word alone."
In order to counter this, Knight suggests the engines should keep a record of search history conducted by specific users and supply results based on previous
searches. Meaning, if the person querying the "thunderbird" keyword was looking for Ford's model and his search history reflected that, the results
could be supplied based on previous contexts.
Of course, relevant results can also be delivered by refining your query as well. In other words, try and be specific when conducting searches and leave the
general keywords alone.
Keep watching SERoundtable and RustyBrick.com for results concerning the
RustySearch relevancy test.