Natural language interaction with automated agents in call centers, with chatbots, and virtual assistants on mobile devices is becoming more and more common. Through developer tools such as Microsoft LUIS, IBM Watson, the Alexa Skills Kit, and Nuance Mix, natural language understanding capabilities are becoming increasingly accessible to developers who want to provide natural interfaces to their applications. In addition, new developments in machine learning such as Deep Learning are leading to rapid improvements in the technology. This workshop covers three topics in natural language interaction. First, we review the spectrum of different types of natural language applications. Then, we look at the technologies that underlie natural language interaction. Finally, we review the natural language understanding development process and work hands-on with some of the current development tools. Attendees are welcome to bring their own ideas for applications for discussion. Bring your laptops.

The workshop provides step-by step guidelines of the speech application tuning process, starting with data collection, data transcriptions, and data tagging and going all the way to delivering a customer-focused recommendation report. Participants gain firsthand understanding of the three aspects of tuning: speech recognition performance, grammar, and dialogue design. The course examines whole call recording tools as well as tools that capture speech event logs and produces metrics. Participants engage in a structured tuning data analysis that targets high-usage dialogue states. In addition to traditional methods, the course introduces an innovative approach to tuning by combining a heuristic usability evaluation with the analysis. This simple, low-cost, low-impact usability assessment provides a caller focus and translates directly to building a business case for the tuning recommendations. Finally, the course explores how tuning methods need to adapt to the broader speech landscape that now includes multi-channel applications, in-car applications, wearables, and mobile devices.

In order to provide a flexible and intelligent service to users, chatbots must be able to understand natural language text and to engage in conversational interaction. The first part of this tutorial introduces a range of natural language understanding (NLU) technologies and highlights those that are particularly relevant for chatbot developers. The second part of the tutorial provides a hands-on exploration of the API.ai tools. API.ai was acquired by Google in 2016 and is being used to provide NLU and conversational capabilities for Google Assistant. Learn how to use API.ai to parse messages into structured data, to predict the next actions to be performed by your bot, and to create slot-filling dialogues. Bring your laptops to develop a sample chatbot.