Daniel BuenoLearning Spanish, Portuguese, and Japanese is the latest stage in the development of Watson, the cognitive computing system launched by IBM in 2011. The Watson platform is able to interact vocally with users in their own language, process large amounts of data, and acquire new knowledge on the fly. “This requires adaptation to the language, including vocabulary and semantic rules,” says Fábio Gandour, chief scientist at the Brazil Research Lab at IBM Brazil. Watson will be “fed” over 300,000 words, as well as being given a processing feature that includes the meaning of each word. “The system is not a finished product that a person can buy and install on a computer or server, it needs to be fed information so that it can produce appropriate responses for each user. It is like a young person who is still evolving,” says Gandour. The new system will enable students, startups, and small companies to design apps for Watson, similarly to those made for Android or for Apple’s iOS. “Watson works with unstructured information. On the websites of banks, for instance, a structured logic prevails, based on predictability and unchanging information. An example of unstructured information would be the result of a Google search for apple pie recipes, which would return more than 150,000 hits. There is no continuous logic among them,” says Gandour. As an additional example, he says that Watson can be a kind of electronic bank manager, indicating the best investments to suit the client’s profile. In hospitals in the USA, where the system is already used, it helps in the diagnosis and treatment of cancer. Some information can be shared among Watson users, but other material can be kept confidential, if so desired by a company that has purchased the system’s services.
Print issue
Current | Past
