Similar to the reading from The Atlantic we had this week, Google along with other companies are trying to organize the world’s data and make them more accessible. With the large growth of data over the recent years, organizing and making sense of it has been a growing field of study. Most of the data mining by the big tech companies has been for targeting advertisements and generating product recommendations. While Google is concentrating on the individual consumer, IBM is focused towards research organizations. As it is unfeasible for a human to read through and cross reference thousands of research papers, IBM aims to solve this by selling Watson’s algorithms as a product.
IBM Watson is known for beating Jeopardy. Watson is a platform that data mines text, stores it and analyzes it. It attempts to find relationships among the data and it sorts the data for fast retrieval. IBM has been working on selling Watson to companies and research organizations as a cloud service. While this may seem a lot like Google’s search engine, IBM is focusing on research organizations and has tuned Watson for mining scientific data. This service can help guide researchers to find: the least researched aspect in a certain field, the right path that will most likely yield new knowledge etc. Many believe that automated mining of scientific literature is very valuable. This makes a lot of sense, as one human can read about 5 papers per day, while Watson can read and analyze thousands of papers per day. A team of researchers have been working on a prototype to further tune IBM Watson for the medical field, namely for their research in cancer. With the help of Watson they were able analyze 70,000 different research papers and figure out which proteins are more likely to mutate to form cancer. IBM’s service was able to identify 7 potential proteins out of the researched thousands. “Some of the things that were predicted turned out to be true, at least in preliminary experiments,” said Dr. Olivier Lichtarge.
In the past year, IBM has largely invested in Watson. It has placed Watson in its own business unit, and IBM is supporting start-ups write modules to further tune the platform for mining data for various specific fields. Whether Watson takes off depends largely on the development side. The original Watson was developed for Jeopardy. Further developing modules for a specific field takes a lot of work. The previously mentioned research team, worked closely with IBM to develop additional software components to allow Watson to understand medical terms. It is yet unclear whether this service will become useful to researchers. However, IBM is exploring a new field where there aren’t any other competitors. Even though, their algorithms are not perfect, with the help of other developers their service will be able to analyze research papers and help researchers across various disciplines.
Article source:
http://bits.blogs.nytimes.com/2014/08/28/questions-for-ibms-watson/?_php=true&_type=blogs&_r=0
No comments:
Post a Comment