When searching the internet, no one really thinks about the underlying technology that makes it all possible. We just want the results. But there are many people working hard to make sure you get those results. One of them is Chris J.C. Burges who is the subject of a new blog post on Technet. He moved to Microsoft Research in 2000 with an explicit aim to work on machine learning projects. By 2005, he had created the basis for the ranking system that is still used in the Bing search engine to this very day.
Named RankNet, the system was ground breaking because it was faster and more accurate than any other system at the time. It could rank results quicker on one PC in a day than any other system could do in several days with multiple computers.
It’s secret? Neural networks; these computer systems loosely mimic the human brain. They can be trained to perform tasks on data labeled by humans. It was unique technology and far ahead of its time. Recently, other search engines and systems have made use of neural networks. They’re used in everything from image captioning to real-time translation.
Burges was drawn to search engine ranking in particular because it was a “hot, highly competitive field in which both researchers and technology companies were competing to be the best.” He stated, “We knew there was a significant opportunity for having an impact.”
Currently, Burges is working on a way to teach machines to read and comprehend text — then have it answer questions about it. Hear him talk about the project in the video below.
It’s a hugely ambitious undertaking that could change the way we interact with machines in the future. Imagine asking Cortana to “read” a book and then asking her questions about it! Burges knows it’s a risky challenge, however, considering his past, he’s relishes taking on long-shot projects he has a passion for.
Further reading: Azure Machine Learning, Bing, Microsoft, Microsoft Research
