Hao Fang. Building A User-Centric and Content-Driven Socialbot. PhD thsis, University of Washington, 2019.

Hao Fang Microsoft Semantic Machines One Microsoft Way, Redmond, WA 98052-6399, USA Email: [email protected] https://hao-fang.github.io Research Interests Conversational AI, Natural Language Processing, Deep Learning Education University of Washington Seattle, WA, USA Ph.D. Electrical Engineering Sept. 2013 – Mar. 2019

Hao Fang (Instructor) – hfang at uw edu; Hao Cheng (TA) – chenghao at uw edu; Course & Office Hours: Lectures: Tuesday 4:30pm – 6:20pm [MEB 237] Labs: Thursday 4:30pm – 6:20pm [MEB 237] Hao Fang: Tuesday 3:00pm – 4:00pm [EEB M406] Hao Cheng: Friday 4:30pm – 5:30pm [EE Atrium] Grading Policy: Labs: 40% (4 labs, each 5% checkoff + 5% ...

Spoken Language Understanding EE596B/LING580K -- Conversational Artificial Intelligence Hao Fang University of Washington 4/3/2018 1

Hao Fang Committee: Hannaneh Hajishirzi Eve Riskin Geoffrey Zweig Mari Ostendorf (Chair) Leah M. Ceccarelli (GSR) Yejin Choi. 1 Agenda o Background o Sounding Board System –2017 Alexa Prize Winner o A Graph-Based Document Representation for Dialog Control o Multi-Level Evaluation for Socialbot Conversations

Hao Fang University of Washington 4/17/2018 Slides adapted from: Andrew Maas, Spring 2017, CS224S/LING285 Spoken Language Processing (Lecture 10&11) Gina-Anne Levow, Spring 2017, LING 575 Spoken Dialog Systems (Lecture 4&5) Content Management 1. Dialog Manager •Takes input from ASR/NLU components

Hao Fang. Outline •Task 1: Create your first Alexa Skill •Step 1: Create an AWS Lambda function •Step 2: Create an Alexa Skill •Step 3: Test your skill using the Service Simulator •Task 2: Create an AWS Lambda function from scratch •Task 3: Deploy the …

How we really talk… A: and thatthat concerns me greatly. B: Well, I don't, -/ yeah, / I'd certainly uhsupport Israel in in their their policy thatin defending themselves and inuhin their handling of their foreign policy, / I thinkI think the stand they have,ororthe way they command respect, II support that./ I think that is aa positive thing for them after um uhthousands of

Q&A vs. Web Search (vs. Task Completion) •When can we use a question answering system? •Answer is known (unambiguously) •Answer is specific entity in the world: e.g. “Space Needle” •When do we need to fall back to web search?

Resources Readings. Spoken Language Understanding: Systems for Extracting Semantic Information from Speech by Gokhan Tur and Renato De Mori; Speech and Language Processing by Dan Jurafsky and James H. Martin. Chapter 29: Dialog Systems and Chatbots Chapter 30: Advanced Dialog Systems Spoken Dialogue Systems by Kristiina Jokinen and Michael McTear, also avaliable at Amazon.