Natural Language Understanding for Customer Service

Interactive Voice Response (IVR) systems are some of the most commercially viable applications of speech recognition. The way people communicate with such systems is advancing rapidly. It is possible for computers to recognize, understand, and respond to conversational input with complicated structured dialogs or limited vocabularies. Periphonics and Unisys recently formed a strategic alliance to automate their customer service processes through intelligent, cost-effective, and time saving methods. The relationship includes a joint development agreement to integrate Unisys’ Natural Language Understanding (NLU) product, the NL Assistant, into the Periphonics IVR platform; a marketing alliance program to increase exposure of Unisys NL Assistant in the Periphonics platform and an OEM agreement, allowing Periphonics to embed the NL assistant into the Periphonics platform and sell the technology to third parties. The NLU platform is intended to extend current environments, according to Unisys, while providing the benefit that comes with a sophisticated interface that includes speech recognition. NLU converts English language input (either spoken words or stored text) into meaning and then responds accordingly. The NLU technology can respond to different versions of the same statement, such as "I don’t know," because it is interpreting the meaning of the words. NLU does more than recognize speech. NLU provides a rules-based context-sensitive technology. Applications can be built to understand a verbal conversation over the telephone, or a typed message over the Internet. With this technology, the NL Assistant can take everyday conversational language, understand what it means, and translate it into computer language. The speech or text is converted into meaning, making interaction with complex computer applications much easier. The goal is to increase customer satisfaction through a more natural user interface. Unisys has been involved in natural language processing since 1981. In 1985, with funding from the Department of Defense Advanced Research Projects Agency, three components developed by Unisys (syntactic, semantic and pragmatic analysis modules) were integrated to form the basis for Unisys natural language software. Seeing the potential for NLU technology led Unisys to begin planning the adoption of this technology for commercial use in 1995. The NL Assistant enhances ASR in two ways: by enabling a more natural dialog and by applying the power of artificial intelligence to extract meaning from the recognizer’s output. NLU can be integrated with continuous speech and speaker-independent ASR technology, whether those systems are speaker-dependent or speaker-independent systems with highly selective vocabularies. Such a combination can enable natural conversation between a customer and a computer.