Watson Discovery, an IBM API adding value to corporate data
IBM’s Discovery API allows companies to structure and understand, with certain ease, the large mass of data they have, in many cases from very different sources, such as documents from office applications, HTML documents, JSON, PDF, etc.
People always talk about data as the key that opens all doors. And they're not wrong. But data is more than obtaining aggregated information (in volume or raw), and companies must be able to generate true business value with it. Achieving this isn't easy. There are a handful of companies that are extracting business value from their data, others are in the process of having aggregated information from which to extract this benefit, and the rest are looking for solutions to have this information; solutions at infrastructure, back and front level.
Whilst there are companies who are searching for this added value, others are offering products and services to these third parties in order to monetize their data. IBM is possibly one of the main companies in the world of artificial intelligence and in the creation of products and services in fields such as processing natural language and machine learning. Most of these products revolve around Watson, the artificial intelligence computer system developed around DeepQA technology (Deep Question Answers). Watson is simply a computer system to search for answers.
This searching for answers is based on the processing of natural language, a computer and artificial intelligence science that studies and improves interaction between machines and human beings, especially at language level. Thanks to this, Watson can understand the world of human beings using senses, learning and people's own experiences. How does IBM Watson resolve questions giving relatively correct answers?
● Watson dives into millions of documents to find answers to the questions it is asked. This is possible because it accumulates large amounts of information from different sources such as Microsoft Word, websites, PDF or JSON documents, documents in cloud applications, etc.
● It uses a scoring algorithm to assess the quality of the answer.
Discovery, an interesting product in Watson
Out of all the products related with Watson, one of the most interesting services is Discovery. The main goal of this product is to help companies easier understand their large quantities of data. This is possible thanks to three situations that are not at all simple in AI:
● Analyzing large amounts of data that may originally be in a format or conditions that are not the most favorable. The information normally comes from completely unstructured sources from local sources as disparate as emails and chats. Most data scientists spend a large amount of time collecting, cleaning and organizing unstructured data. It is arduous, complex and necessary work to obtain information with real value.
● IBM Watson is also able to analyze structured and semi-structured information from other sources such as databases or the content management systems (CMS) that companies tend to have.
● Artificial intelligence is normally able to give one determining response to a question or query. The really difficult aspect is for a system or platform to be able to contextualize and provide alternatives. The goal of Watson Discovery is to contextualize data and provide responses.
The Watson Discovery API
IBM boasts a service that doesn't need a PhD to be able to use it. With know-how, a call to a REST API is necessary. This is the way in which the entire process of extracting, transforming and loading (ETL) content is carried out and directly integrated with Watson's capabilities of importing documents and large volumes of data. From there, developers can identify tasks, analyze data, explore information, search, etc.
In the end, Watson's large block of APIs allows users to:
● Track, convert, enrich and normalize data.
● Explore the content of the client as well as from totally free public content sources, or content that is under license.
● Add a new value to the information collected: Watson uses natural language processing to extract concepts, establish relationships and link feelings to data that has been collected.
● Simplify development while still providing access to APIs.
The goal of Discovery is to put all the capabilities that IBM Watson has in technologies, such as computing and natural language processing, into a service that is more accessible to all, with the idea of supporting all types of sectors and contents, especially those that handle large amounts of data such as health. With Discovery, the idea is to make it very simple to prepare unstructured data, make the necessary queries to locate the information the user wants and integrate all that into a new application, a platform or a solution.
How content is added with Watson Discovery
Watson Discovery has three different ways of relating, contextualizing, connecting, and finding brilliant ideas from among all the data that a company has after the process of tracking and organizing the unstructured information:
● Using the API: this is the best possible formula if the content loading is directly integrated with an existing application or service.
● Discovery tools: the ideal method if the user wants to directly upload this content to a local service for querying and using.
● Using the data tracker: the data tracker is a command-line tool that helps developers take their documents from the repositories they are in (for example a database) and push them to the cloud to be used by Watson Discovery.
If you want to learn more:
Sign up to the BBVAOPEN4U newsletter and receive tips, tools and the most innovative events directly in your inbox.