How search engines work today
Search engines today are not very smart. I say this because, face it, many searches do not return exactly what we want.
They offer a seemingly random list of related content, that’s usually not on target. We often have to refine and re-enter our search.
In a paper published in the Journal of Information Science by A. Uyar, multi word queries can affect search results accuracy by 50%. And as the number of words increase, the accuracy drops.
Search engines use a number of factors to rank websites. In many cases this method fails.
This is partly the fault of the user as well. The user often enters vague keywords that the engines cannot properly interpret. This issue is not handled gracefully.
The problem with the current search engine model
The problem is Search Engines do not know what a user mean or intend in their query.
The search engines are now experimenting with live suggestions to help the process. But this method must be refined.
Another problem is search engines don’t always rank sites based on relevance to search. They try to make a guess at what is really expected from the request, and offer the best based on their very imperfect ranking methods.
Ideas that might help improve search results
There are a number of research projects currently trying to tackle this issue.
For example XFN (XHTML Friends Network). Using hyperlinks we are able to represent human relationships. Website owners can state who they are linking to and how they know that person or site.
Also FOAF (The Friend of a Friend), with their method we are able to do similar relationship declarations as XFN, but with more control over information. And with more access to deeper attributes and connections of each relationship.
There is also Hakia. This is a Semantic Search Engine. Their method is based on the idea of more credibility than popularity. They use librarians to judge the content among other factors.
These are just three examples of the next generation of technology that will make search engines just a little more intelligent.
How these can be implemented
Implementing some of these new technologies might not be so practical. For one, there needs to be far more options available through XFN and FOAF. The Hakia search engine uses humans, and this might not be easy to scale.
There are other projects that might prove more reliable than presented here. But currently no one person or company has made significant strides to building something remotely as smart as a child.
How will the future of search engines look
The future of search might prove to be revolutionary. The right information will be instantly available in a streamlined and natural way.
For example, we speak our query and have the request taken by our in house computer system, and then give us the answer in voice, video, text or any number of other formats. This would of course be highly intelligent and would return only the most relevant and credible results.
The search engines are not smart enough. Their ranking methods are faulty and need serious upgrading.
There are technology being researched and developed that might help. There exists the problem of scaling because the amount of information required cannot be handled effectively at the moment.
In the future, search will be radically more intelligent. They will give almost perfect answers to our questions, instantly and in any format we desire.