Testing Google Personalized Search
Let's face it. Google was an innovating search engine that changed the way people thought about searching. I've recently been remembering the older search engines I used to use seven or eight years ago: Dogpile, Metacrawler, AskJeeves, and even Yahoo!. That was back when searches used to take tens of seconds rather than tenths. Ah, "The good ol' days."
Just kidding.
Google introduced us to speed, quality, and a simple interface. Thank you, Google.
However, searching still isn't perfect (not for lack of trying, I'm sure). Ideally, I could ask a question, the search engine would know what I was talking about, and it would know exactly what page or pages to direct me to.
Except... how can a machine know what I am thinking, in what context my question is being asked? That is what my research is investigating - how to inform a search engine the context of a question.
Google already has some functionality for doing this. Google Personalized Search keeps track of your search history and hypothetically adjusts your search results to help narrow your search to topics that have interested you in the past.
However, my testing has not been successful in narrowing my search.
I started with a simple search query: jaguar
There are three distinct categories of results: Jaguar the car, Jaguar the Operating System (MAC OS X v10.2), and Jaguar the feline animal. Jaguar the car was the top result, followed by Jaguar the OS, then a single Jaguar the animal.
I attempted to promote Jaguar the OS to a higher rank. To do this, I deleted my entire search history, then began building the profile I wanted. I searched for Macs, I clicked on search results that had to do with Macs, I did everything I could think of to tell Google that I was very interested in Macs.
Then... I searched for jaguar again. Unfortunately... it returned the same results as before, in the same order. So, something is not working...
Just kidding.
Google introduced us to speed, quality, and a simple interface. Thank you, Google.
However, searching still isn't perfect (not for lack of trying, I'm sure). Ideally, I could ask a question, the search engine would know what I was talking about, and it would know exactly what page or pages to direct me to.
Except... how can a machine know what I am thinking, in what context my question is being asked? That is what my research is investigating - how to inform a search engine the context of a question.
Google already has some functionality for doing this. Google Personalized Search keeps track of your search history and hypothetically adjusts your search results to help narrow your search to topics that have interested you in the past.
However, my testing has not been successful in narrowing my search.
I started with a simple search query: jaguar
There are three distinct categories of results: Jaguar the car, Jaguar the Operating System (MAC OS X v10.2), and Jaguar the feline animal. Jaguar the car was the top result, followed by Jaguar the OS, then a single Jaguar the animal.
I attempted to promote Jaguar the OS to a higher rank. To do this, I deleted my entire search history, then began building the profile I wanted. I searched for Macs, I clicked on search results that had to do with Macs, I did everything I could think of to tell Google that I was very interested in Macs.
Then... I searched for jaguar again. Unfortunately... it returned the same results as before, in the same order. So, something is not working...
- I haven't put enough time into building my Mac profile
- Google is unable to see the similarity between the sites in my search history and the sites that are returned when I search for Jaguar
posted by Alias:Kilroy at 12:00 PM
0 Comments:
Post a Comment
<< Home