IBM Adds NLP Tools to Watson AI
Source: IBM
IBM has upgraded its Watson AI platform to boost the ability to understand business lingo that could be used to analyze the English language text and conversations.
The natural language processing capability represents the first commercialization of IBM Research’s Project Debater, billed as an AI system capable of debating humans. The tool is designed to help develop arguments that could ultimately be used by support decision making.
IBM said this week it is bringing the fruits of the research project to the enterprise market to improve machine understanding and help analyze complex human speech.
Among the challenges in developing AI platforms that can understand conversational English is training them to understand colloquialisms and idioms. The trick is to develop a system that can interpret a phrase like “You hit the nail on the head!” as “You are exactly right.”
IBM said Wednesday (March 11) its Watson NLP capability also could be used to parse contracts, PDFs and other business documents. The integration of Project Debater technology with Watson includes a sentiment analysis and “summarization” tools along with the ability to cluster data by topic.
The sentiment analyzer being released this week can, for example, detect “sentiment shifters,” so-called because a negative modifier can reverse a speaker’s meaning using a phase like “hardly helpful.” The analysis tool would also enable the Watson API to identify common idioms like “calling it a day.”
The briefing tool extracts text data from different sources to provide a summary of speech and text on a specific topic. IBM said the summarization tool was used during this year’s GRAMMY awards and will be integrated with its Watson natural language understanding platform later this year.
The clustering feature allows users to organize data and generate topics. Available later his year on IBM’s Watson Discovery enterprise search platform, the data organizer is aimed at subject experts in specific industries like health care, insurance and manufacturing.
The NLP platform also includes a tool for creating AI models that can be used to classify specific clauses in procurement contracts and other business documents. The classification tool is based on deep learning technology developed under Project Debater. IBM said the tool can be trained using as few as several hundred samples.
Project Debater was introduced as part of a live debate with humans in June 2018. The NLP technology demonstrated the ability to construct an argument, consider counter-arguments and deliver a rebuttal.
“Language is a tool for expressing thought and opinion, as much as it is a tool for information,” said Rob Thomas, general manager of IBM’s data and AI unit. “Advancing our ability to capture, analyze, and understand more from language with NLP will help transform how businesses utilize their intellectual capital that is codified in data.”
IBM is now focusing its NLP research on classification, retrieval and speech generation, Sriram Raghava, vice president of IBM Research AI, added in a blog post announcing the Watson tool.
