IBM Watson and Natural Language Processing

IBM Watson beating Jeopardy champions
IBM Watson beating Jeopardy champions

The computers we use today are built using mathematical models, which are created using approximations that allow us to simulate realities. However, a majority of the data in the world today, approximately 80%, is what is called “unstructured data” which means it cannot be processed by computers. This includes text, literature, audio and video. This is because today’s computers are dependent on being programmed to perform finite tasks and are unable to ingest and understand the context of human language.

IBM Watson was created to resolve this problem – Watson is a super-computer that understands language and as a result the cognitive process. Our cognitive process as humans is fundamentally based on our understanding of language, which is how we contextualise our experiences and draw meaning from the world around us. IBM Watson is at the forefront of this emerging field called cognitive computing, which is developing systems rooted in computational linguistics as opposed to mathematical models. Researchers are teaching Watson to understand natural language, which in turn leads to cognition and the capacity for it to learn. With the emerging field of cognitive computing and IBM Watson we’ll soon start to experience real applications of Artificial Intelligence (AI) and a revolution in human-computer interaction. In November 2013, IBM launched a cloud-based API service, providing developers with access to Watson’s computing capabilities. Developers across all sectors, both commercial and non-profit, are now starting to prototype and incorporate the power of cognitive computing into mobile apps and software.

UX Researchers in IBM’s Visioneering team have developed and are evolving a distinct personality for IBM Watson – the UI uses natural language processing to understand and learn the context of spoken and written language. In other words, IBM Watson reads and interprets text like a person, it does this by breaking down a sentence grammatically, relationally and structurally – discerning meaning from the semantics of the material. Watson understands context, this is different than voice recognition software, which is how a computer handles speech. Watson tries to understand the intent of the user’s language and uses that understanding to extract logical responses and draws inferences to potential answers through a broad array of linguistic models and algorithms.

Watson’s personality (or UI) was developed by IBM’s Visioneering team through “UI exploration and research – with the aim to identify the ideal interaction between human and computer.” The use of natural language as the user interface for Watson wasn’t motivated by aesthetics or novelty but rather it is considered a powerful tool in human-computer interaction – conveying information, requests and constraint in an efficient manner. Users are able to communicate with Watson using natural language, such as without having to break down commands or questions into smaller sub-tasks. This intelligent interface understands a user’s higher goals and is able to fill in the blanks without implicit information.

Over time Watson is able to adapt and learn from the user’s selections and responses – providing users with real-time personalisation within the interface.

In the near future, it is likely that natural language processing and interfaces will becoming increasingly ubiquitous – transforming the way we interact with computers.