Keyword Search Is Not Productive. Who Says?
June 30, 2015
I noticed a flurry of tweets pointing to a diagram which maps out the Future of Search. You can view the diagram at http://bit.ly/1LznR1U or http://bit.ly/1dugOcR. Direct your attention to this assertion:
As amount of data grows, keyword search is becoming less productive.
Now look at what will replace keyword search:
- Social tagging
- Automatic semantic tagging
- Natural language search
- Intelligent agents
- Web scale reasoning.
The idea is that we will experience a progression through these “operations” or “functions.” The end point is “The Intelligent Web” and the Web scale reasoning approach to information access.
Interesting. But I am not completely comfortable with this analysis.
Let me highlight four observations and then leave you to your own sense of what the Web will become as the amount of data increases.
First, keyword search is a utility function, and it will become ubiquitous. It will not go away or be forgotten. Keyword search will just appear in more and more human machine interactions. Telling your automobile to call John is keyword search. Finding an email is often a matter of plugging a couple of words into the Gmail search box.
Second, more data does translate to programmers lacing together algorithms to deliver information to users. The idea is that a mobile device user will just “get” information. This is a practical response to the form factor, methods to reduce computational loads imposed by routine query processing, and the human desire for good enough information. The information just needs to be good enough which will work for most people. Do you want your child’s doctor to take automatic outputs if your child has cancer?
Third, for certain types of information access, the focus is shifting, as it should, from huge flows of data to chopping flows down into useful chunks. Governments archive intercepts because the computational demands of processing information in real time for large numbers of users who need real time access are an issue. As data volume grows, computing horsepower is laboring to keep pace. Short cuts are, therefore, important. But most of the short cuts require on having a question to answer. Guess what? Those short cuts are often keyword queries. The human may not be doing keyword searching, but the algorithms are.
Fourth, some types of information require both old fashioned Boolean keyword search and retrieval AND the manual, time consuming work of human specialists. In my experience, algorithms are useful, but there are subjects which require the old fashioned methods of querying, reading, researching, analyzing, and discussing. Most of these functions are keyword centric.
In short, keyword queries can be dismissed or dressed up in fancy jargon. I don’t think the method is going away too quickly. Charts and subjective curves are one thing. Real world information interaction is another.
Stephen E Arnold, June 30, 2015
Comments
One Response to “Keyword Search Is Not Productive. Who Says?”
-
- Subscribe to Beyond Search
Feature archive
News archive
-
I can’t bear reading long posts, simply because i have got a small amount of dislexia, but i actually enjoyed this one