Indian-origin 16 year old claims his search engine is 47% more accurate than Google
 |
| Anmol Tukrel |
Watch out Sundar Pichai. There's a high school project that tops Google's main cash cow, the search engine.
Sixteen-year-old Anmol Tukrel, an Indian-origin Canadian citizen has designed a personalized search engine that claims to be as high as 47% more accurate than Google, and about 21% more accurate on an average.
Tukrel, who just completed his tenth grade, said he took a couple of months to design it, and about 60 hours to code the engine, as part of his submission into Google Science Fair, a global online competition that is open to students aged 13 to 18 years.
"I thought I would do something in the personalized search space. It was the most genius thing ever. But when I realized Google already does it, I tried taking it to the next level," said Tukrel, who was in India for a two-week internship programme at Bengaluru-based adtech firm IceCream Labs.
Tukrel's tinkering kit: A computer with at least 1 gigabyte of free storage space, a python-language development environment, a spreadsheet program and access to Google and New York Times.
To test the accuracy of each search engine, he limited his search query to this year's news articles from The New York Times. He created several fictitious users with different interests and corresponding web histories. Tukrel then fed this information to both Google and his interest-based search engine. Finally, the results from each search engine were compared.
Sixteen-year-old Anmol Tukrel, an Indian-origin Canadian citizen has designed a personalized search engine that claims to be as high as 47% more accurate than Google, and about 21% more accurate on an average.
Tukrel, who just completed his tenth grade, said he took a couple of months to design it, and about 60 hours to code the engine, as part of his submission into Google Science Fair, a global online competition that is open to students aged 13 to 18 years.
"I thought I would do something in the personalized search space. It was the most genius thing ever. But when I realized Google already does it, I tried taking it to the next level," said Tukrel, who was in India for a two-week internship programme at Bengaluru-based adtech firm IceCream Labs.
Tukrel's tinkering kit: A computer with at least 1 gigabyte of free storage space, a python-language development environment, a spreadsheet program and access to Google and New York Times.
To test the accuracy of each search engine, he limited his search query to this year's news articles from The New York Times. He created several fictitious users with different interests and corresponding web histories. Tukrel then fed this information to both Google and his interest-based search engine. Finally, the results from each search engine were compared.
 |
| Anmol Tukrel & Sundar Pichai |
Today, personalization is dependent on factors like one's location, browsing history, and the affinity to the kind of apps they install on their phone. That's just one part of the equation. Tukrel claims his algorithm solves the other side of the equation: It understands what a user would like before it serves up the results by dwelling deep into the content of the text, understanding the underlying meaning, before matching it to a user's personality, and throwing up the result.
"For someone to look at a successful Google product and attempt to go one level up, it's astonishing," said Sanjay Ramakrishnan, co-founder of Ice-Cream Labs, and former marketing manager of Myntra. Tukrel, the student of Holy Trinity School in Toronto, said he learnt to code in his third grade, and subsequently picked up on mathematics and coding.
"My computer teacher was pretty impressed with the project. I skipped a year in computer science, so they knew I was good, but may be not so good," said Tukrel, who has put up a link to the test cases online for anyone to view.
Tukrel submitted his paper to the International High School Journal of Science last month, and hopes to study computer science at Stanford University. But before that, he wants to develop a news aggregator based on this technology, and licence it to a few digital marketing agencies as well. Would he become a fellow at Paypal founder Peter Theil's foundation, where one is required to drop out of college to try an idea?
"To be honest, it's incredibly stupid to drop out," said Tukrel. "It's very arrogant to think that your idea is so good, that you don't need to learn anything."
But, Tukrel also runs a company, which has a palindromic name: Tacocat Computers. But, is he legally allowed to?
"Yes. You just need parental consent."
And, what next?
"Eleventh grade."
Courtesy : TOI
wow. Great.
ReplyDelete