6

The ability for a computer system to understand natural language—the unstructured written information that accounts for the bulk of information produced and consumed by humans—is a significant breakthrough. For the first time, machines can comprehend and extract semantic knowledge from any unstructured data resource, including: Research papers Technical, business and news articles Books and essays Personal blog articles, reviews, tweets and social media posts Images and photos To interpret information like a person, the cognitive computer breaks down each sentence into its syntactic structure. It looks at the relationships of words and phrases and their context within the whole document. It then looks for relationships with other words and phrases from other documents available to it. This ability to absorb knowledge like a human helps the computer answer human questions, posed in natural language, with truly human-sounding answers. The objective is to understand the intent of the user’s question and use that understanding to develop logical responses, draw inferences, generate potential answers, and settle on the best response backed up with source information. Learning the language unique to each industry When someone begins working in a new industry, the first job is to learn the language specific to that industry. In specialized fields, practitioners even have to learn the thought patterns that are unique to it. In medicine, for example, there are multiple domains, each with its specialized language and knowledge base. 6