Conversational user interfaces and natural language processing could be put to much more use than they currently are. At the GigaOM Structure Data event in New York City this week, IBM distinguished engineer Currie Boyle, who leads the vendor’s North American natural language services practice, including for deep question and answer Watson-type natural language and unstructured information processing systems, and Nuance Communications CTO Vlad Sejnoha, discussed the realized promises, but also the waiting opportunities.
At Nuance, Sejnoha noted, the focus is on the notion that we are entering a time when how we interact with systems and access information and content is undergoing a “dramatic transformation.” Contributors to that include high- level artificial intelligence reasoning and natural language understanding. “We are overwhelmed with lots of data including unstructured data and these technologies make a difference in how we take advantage of all that,” he said.
“This ecosystem change is happening in the industry, Boyle concurred, discussing the desire for business dialogue management systems to try to determine the intent of a user seeking information and the intent of the author who wrote it, and matching the two by that intent, even if they don’t share the same words in common to express it. The applications range from consumer conversational and context-aware systems to business professionals finding answers in structured or unstructured data through via natural language interfaces to boosting call contact center performance with dialogue management.
“The transformative dimension starts with understanding the intent of text or the spoken word, extracting meaning from it and then determining the next best action and down the line engaging in collaborative dialogue,” Sejnoha said. That requires disambiguation of content, and weighing of different sources of information, and both companies are building the full stack of technology to support this. “This is moving toward reasoning, toward AI, where you encode the overall goals and the system dynamically determines how to best advance the transaction interacting with both stuctured and unstructured data.” he said.
But one issue to overcome to transform the user experience in this direction is broadening the content sources — the other services and applications — with which such conversational stacks will be able to interact. “Today integrating these interfaces to those resources is one-off,” said Sejnoha. “That opens an important question about openness as we evolve the UI for natural language and direct access to specific desired [information]’”
And that’s where the promise of the semantic web remains very important, he said. “I’m hopeful to get to the point where people with important content or services on the web publish in standard formats, … so that we connect to them and bring things in automatically.” The W3C’s semantic web standards, he said, hopefully will gain greater support, or else integrating new content services into conversational virtual assistants is more difficult. “Connecting natural language systems to new content is not trivial but it is aided by standardizing the description of resources,” he says.
Moving towards a conversational natural language interface as part of any application’s foundation also needs to be a bigger part of the conversation among the software community, Boyle noted. It’s time to start asking how it’s possible to differentiate one’s application by working with vendors such as Nuance, IBM, and others to offer speech and text-based natural language systems to do things more effectively. “That’s available today,” he said. “But a lot of people we talk to have a limited point of view of what’s doable and some organizations can deliver more than what the industry expects in terms of understanding intent and turning it into an action.”
- MOLTO: Improving Online Text Translation with Machine Learning
- The Delectable, Semantic Future of Whisk
- Watson is Getting Smarter
- Analytics for Musicians with BeatDeck