Introduction to IBM’s Watson
IBM Watson may be famous for taking part in competitions on the US quiz show Jeopardy. However, the technology is not all about answering simple questions. Watson is a whole new form of computing, which has huge potential to discover new insights based on the amount of data found around us. The major difference between Watson and conventional computer models is that Watson relies on what IBM refers to as “cognitive computing,” while the conventional computers relied on rigid mathematical principles, using software built on logic and rules.
Cognitive computing is different from traditional computing because it lets Watson comprehend unstructured data from a variety of sources. Computers, generally, like to work with structured data that are, for instance, contained in a database. However, with the rise of big data, it has become impossible to structure the huge amount of data being created per second. As a matter of fact, itis estimated that 90% of all digital data is unstructured. Thus, IBM Watson recognizes that computing must adapt to this new data set to generate the kind of valuable insights that will be beneficial to the world.
Therefore, Watson is defined as an IBM supercomputer, which combines artificial intelligence with sophisticated software for best performance. The computer is named after the founder of IBM, Thomas J. Watson.
Watson processes at a speed of 80 teraflops (trillion floating-point operations per second). Watson has the capacity to duplicate or exceed a high-functioning ability of a human to answer questions by accessing 90 servers with a combined data store of more than 200 million pages of data. Watson then processes these data against 6,000,000 logical rules. The Watson super-computer and its data are housed in a space that is large enough to accommodate ten refrigerators.
Watson’s key components are:
- Apache Unstructured Information Management Architecture (UIMA) frameworks, infrastructure and other elements needed to analyze unstructured data.
- Apache’s Hadoop, which is a free Java-based programming framework supporting the processing of huge data sets in a distributed computing environment.
- The fastest available Power7 processor operating system (SUSE Enterprise Linux Server 11).
- 2,880 processor cores
- 15 terabytes RAM
- 500 gigabytes of preprocessed data.
- IBM’s DEEPQA software that is designed for data retrieval incorporated with natural language processing and machine learning.
Watson can understand unstructured data since it uses a vastly different approach to information than the conventional computers. Natural language has a challenge since it involves implications, ambiguity, and idioms, and it is mostly defined by its context as well as its content. Nevertheless, Watson can match keywords in unstructured data and also interpret the text. It does this by breaking down sentences into grammar, relation, and structure. Watson tries to understand the true intent of the language so that it can respond logically and deduce insights.
With regards to the above, Watson’s method to information can be compared to that of human, but in this case, using technology to harness the same procedures of observation, making an interpretation, carrying out an evaluation, and making a decision. It, therefore, means that Watson cannot only follow instructions but can also learn. Wherever Watson is employed. It learns the relevant context, language and thought process previously from human experts. By furnishing Watson with the relevant body of literature, it can develop an amount of knowledge within a particular field. Watson then receives training through a process called machine learning, in which pairs of questions and answers give Watson the basis to interpret the huge amounts of data available to it.
Currently, over three hundred partners are using Watson to develop cognitive solutions, and some of them are already on the market. Anyone can access Watson on a simple pay per use model through the IBM’s public cloud development platform, Bluemix. There is also the option for Shared risk models for complex projects.
IBM Watson is already finding its usefulness across different industries. For example, in healthcare, Watson has been used to analyze patient medical history, with the help of existing data, to suggest more effective, practical treatment options for cancer patients. For the time being, Watson analytics has already proved its usefulness in the marketing industry, sales, operations, customer service, human resource, and finance. The potentials of Watson are limitless, and so the most important thing may not be to question what Watson is capable of doing now, but what it is capable of in the future. Watson is different from other computers because it can learn, adapt, and keep getting better and smarter.