Information retrieval through conversation

Over the weekend I read an article on Slate by the great Farhad Manjoo (a writer for Slate) discusses Google chasing the vision of creating a search engine like that used on Star Trek’s USS Enterprise.  The article said:

The Star Trek computer worked completely differently. It understood language and was conversational, it gave you answers instead of references to answers, and it anticipated your needs. “It was the perfect search engine,” Singhal said.  [Amit Singhal is senior vice president and software engineer at Google Inc., a Google Fellow, and the head of Google’s core ranking team. ]

It struck me that now search for information is only single-query – we only get to ask one question and get one set of data in return.  Could it be possible to narrow down our hunt for knowledge with multiple queries?  It seems likely in the future that search is going to become multi-query.  Just imagine if we could break up a query into different sets so as to narrow down the result till it finds exactly what we want.  Perfect.

When we go into a book store to search the inventory and make a purchase we start with the simple search for a “book”.  From there we start narrowing our search down even further to, say, “travel books”. From there we might search either “fact or fiction” or perhaps a geographical area, like Asia. We would then look for our detailed topic — a factual travel book about Thailand, for instance.

This is how decision making happens when we know what we want, and need to drill down to find what are looking for.  This is very simple because the outcome is a discrete object and the process is extremely logical.

When we go to a restaurant we also use a fairly logical process to find what we want to eat.  We start by making some big decisions before we leave the house: what kind of food do we want to eat? How much do we want to spend, and how much time do we want to dedicate to our evening out?

These are just a couple of examples of processes that are very familiar and easy.  Searching for books and a nice meal out is a logical process.  But there are many decision-making processes that are new to us. These pop up as we go about acquiring knowledge and some are quite complicated.  Just imagine we have a process that ends with choosing two very different solutions but neither is perfect.

In such cases what we might need is more like a guide — a guide that not only gives us answers to our questions but also,  when we are unsure what we are looking for, proposes further questions.

Wouldn’t it be great if the search engine could ask questions of the user when the user is unable to narrow down the results?  If a user were looking for books about Game Theory, for instance, but knew nothing about Game Theory, the user would have no way to narrow down the search to a specific topic or source.

I think the ideal search engine is a guide to knowledge and not simply an information retrieval system.  It needs to be able to give answers, and also prompt users with the right questions.

But how do we integrate this type of searching into our own websites? When looking at this from the perspective of someone who builds websites and strategy for large commercial websites, I see nothing but advantages in this expanded way of thinking:

  • We need to create great content that provides answers.
  • We need to provide answers to multiple user levels – such as beginner, intermediate, or advanced.
  • We need to realize that websites must have structures that allow users or robots to understand how information contained on the website is organised.
  • We need to make sure that every piece of content is helpful to a user in some way.
  • Content should leave the user fulfilled, and that would be defined by having provided a good answer to the user’s query.

Leave a Reply

Your email address will not be published. Required fields are marked *