Google Copies from Ask.com
March 25, 2009
Newsvine.com ran a story bylined by Michael Liedtke, a journalist working for the Associated Press. I am fearful of quoting anything from an AP story, but I think I can convey the gist of the story “Google Draws upon Rival Ideas with Search Changes” here. The idea is that Google’s suggested queries were inspired–Mr. Liedtke uses the word “popularized”–by Ask.com, the search engine of NASCAR. When I read this, I laughed. Suggested searches are not exactly a new innovation. I looked in my files and found references to clustering dating back a decade or more. I recall a clustering effort coded by the Information Industry Award recipient Howard Flank in 1981. The difference between the early attempts at clustering and what Google introduced boils down to one word–scale. Mr. Flank’s effort would not run on the machines available to us at the proprietary Lockheed Dialog company. My thought is that the Google has a practice of working through innovation from computer labs and research papers, learning, and using its clever methods to implement useful functions on Google’s scale. Was Ask.com the inspiration for Google’s engineers. A more likely influence was Dr. Salton’s 1978 paper “Generation and Search of Clustered Files”. Need a copy. Click here. I have zero relationship with Googzilla, an outfit wishing I was a roasted goose. But I was taken aback with the suggestion that the GOOG turned to the search engine of NASCAR for inspiration.
Stephen Arnold, March 25, 2009
Comments
3 Responses to “Google Copies from Ask.com”
-
- Subscribe to Beyond Search
Feature archive
News archive
-
I’ve always admired the Ask 3D interface and how it populates (and clutters) every inch of the screen with data fuzzily related to the query, no matter how vague it may be.
A couple of years ago I had the privilege of playing with the Grokker product by Groxis which introduces similar functionality but presents it in a graphical, manipulable fashion which provides a wonderful sense of relationship and scale of correlated data points to the end user.
Soon after Google introduced results clustering on their Google Search Appliance which does a really astounding job of generating fairly insightful recommendations, even on one-word queries, in real time based on the actual contents. No matter how many (or few) documents there are in the index the GSA produces these clustered query recommendations in no more than about 4 seconds. Your mention of “Google’s scale” is what prompted me to think again of how amazing this functionality is, since the GSA is but one box! Sometimes scale can be the innovative differentiator, even when the technology being scaled is a quarter-century old.
By the same reasoning – search in nothing else but information retrieval. Nothing is new – everything was invented in the sixties and seventies, yes?
You are probably an old foggy fella – who read his last paper in 1981. Technologies which ask proposed first are far more intricate than just clustering – dealing with specific corpus of user queries. It is not just “scale”.
Kwaldrania,
Yep, I haven’t read or done anything since 1981. Your acumen underscores your knowledge and research skills. You are a genius,
Stephen Arnold, March 27, 2009