Internet web searching still uses what is basically 1990s full-text searching technology. It’s time for an update.
1990s technology took every word in a search query and tossed out noise words like “of” and “the”. It then had a list of documents with those words. You would do an intersection of the lists for all search query terms. The more advanced ones had a “word count” so you could rank order the results. That’s about it.
The only thing google does (did) was they figured out how to do it on very large sets of documents. Basically, they developed sparse boolean sets so they could handle an intersection of a billion documents. But that is behind the scenes. You still get the same old results.
What Google lacks is any real understanding of what you are asking for. If I ask it to search “All BSP documents on #1A ESS issued after 1975” Google is utterly clueless. It spits back documents with 1975 in them, and documents for the 1ESS and not only the 1AESS. In short, Google is an idiot savant.
When will Google start to understand ideas, and sentences and strong word associations? When can I ask it for answers to questions. And, above all else, when can I have it stop giving me links to talk forums filled with chatter from uninformed people asking off-topic questions? It needs to understand that not all documents have equal credibility.
When I saw IBM’s Watson I immediately exclaimed that it was the next Google. Indeed, it could be. It should be. I want it to be.
Watson understands meanings and sentences and prioritizes contents.
Whichever company brings a Watson-like company out for the public to use is going to drive Google out of business. Everybody will switch. It will kill Google.
Oh, wait, no it won’t. We no longer have enforced anti-trust laws. Google will just buy the company. They have a lot of cash. Then, the MBAs will do a business plan and determine it is cheaper to just shelve the company than to upgrade all of the Google search infrastructure and risk disrupting the business. Google could do this repeatedly and it wouldn’t total 10% of its cash. Who knows, perhaps they are already doing this strategy.