Master of Science / Department of Computer Science / William H. Hsu / This report presents a personalized exercise training chatbot for individual users based on data collected from the Internet of Things (IoT), particularly wearable fitness devices. The chatbot is designed with our goal of motivating users to exercise more by discussing exercise statistics with the user, such as whether their daily steps have increased, decreased, or remained steady. In this work I first survey a few examples of how increased interest in fitness and the promotion of healthy lifestyles is driving demand for personalized artificial intelligence, wear- able computing, and ubiquitous computing applications. Next, I describe the design of a data-driven ”personal trainer” chatbot. I then develop a prototype persuasion system based on interactive dialogs delivered via a front-end application, that collects data from wearable equipment using back-end data loggers that I instrumented as a mobile application. Finally, I describe the process of deploying and demonstrating this prototype along with technical challenges and early findings. The overall system consists of (1) the back-end Coach agent, an Android application that collects data from all wearable instruments, and (2) the front-end Me agent, which initiates and continues conversations with the user using notifications that are in turn based on data from the Coach agent. This data-driven ensemble reminds the user to exercise and also gives the user a chance to provide feedback via human/agent interactive dialogs. In this project, I used only one wearable device, the MI Band 2, and get real-time steps and weekly step aggregates from it. The human/agent dialogues are deployed via the Slack groupware platform. Google Sheets is used as a web service for updating and exchanging data.

In this exciting era of cognitive computing, conversational agents have a promising utility and are the subject of this thesis. Conversational agents aim to offer an alternative to traditional methods for humans to engage with technology. This can mean to reduce human effort to complete a task using reasoning capabilities and by exploiting context, or allow voice interaction when traditional methods are not available or inconvenient. This thesis explores technologies that power conversational applications such as virtual assistants, chatbots and conversational agents to gain a deeper understanding of the frameworks used to build them. This thesis introduces Foodie, a conversational kitchen assistant built using IBM Watson technology. The aim of Foodie is to assist families in improving their eating habits through recipe recommendations taking into account personal contexts, such as allergies and dietary goals while helping reduce food waste and managing grocery budgets. This thesis discusses Foodie's architecture and derives a design methodology for building conversational agents. This thesis explores context-aware systems and their representation in conversational applications. Through Foodie, we characterize the contextual data and define its methods of interaction with the application. Foodie reasons using IBM Watson's conversational services to recognize users' intents and understand events related to the users and their context. This thesis discusses our experiences in building conversational agents with Watson, including features that may improve the development experience for creating rich conversations. / Graduate

Cybergrooming refers to the crime of establishing personal close relationships with potential victims, commonly teens, for the purpose of sexual exploitation or abuse via online social media platforms. Cybergrooming has been recognized as a serious social problem. However, there have been insufficient programs to provide proactive prevention to protect the youth users from cybergrooming. In this thesis, we present a generative chatbot framework, called SERI (Stop cybERgroomIng), that can generate simulated conversations between a perpetrator chatbot and a potential victim chatbot. To realize the simulation of authentic conversations in the context of cybergrooming, we take deep reinforcement learning (DRL)-based dialogue generation to simulate the authentic conversations between a perpetrator and a potential victim. The design and development of the SERI are motivated to provide a safe and authentic chatting environment to enhance the youth's precautionary awareness and sensitivity of cybergrooming while any unnecessary ethical issues (e.g., the potential misuse of the SERI) are removed or minimized. We developed the SERI as a preliminary platform that the perpetrator chatbot can be deployed in social media environments to interact with human users (i.e., youth) and observe the conversations that the youth users respond to strangers or acquaintances when they are asked for private or sensitive information by the perpetrator. We evaluated the quality of conversations generated by the SERI based on open-source, referenced, and unreferenced metrics as well as human evaluation. The evaluation results show that the SERI can generate authentic conversations between two chatbots compared to the original conversations from the used datasets in perplexity and MaUde scores. / Master of Science / Cybergrooming refers to the crime of building personal close relationships with potential victims, especially youth users such as children and teenagers, for the purpose of sexual exploitation or abuse via online social media platforms. Cybergrooming has been recognized as a serious social problem. However, there have been insufficient methods to provide proactive protection for the youth users from cybergrooming. In this thesis, we present a generative chatbot framework, called SERI (Stop cybERgroomIng), that can generate simulated authentic conversations between a perpetrator chatbot and a potential victim chatbot by applying advanced natural language generation models. The design and development of the SERI are motivated to ensure a safe and authentic environment to strengthen the youth's precautionary awareness and sensitivity of cybergrooming while any unnecessary ethical issues (e.g., the potential misuse of the SERI) are removed or minimized. We used different metrics and methods to evaluate the quality of conversations generated by the SERI. The evaluation results show that the SERI can generate authentic conversations between two chatbots compared to the original conversations from the used datasets.

During the past few decades, there has been increasing attention to multimodal adaptive language learning interface design. The purpose of this study was to examine users’ experiences with a chatbot language learning interface through the lens of cognitive emotions and emotions in learning. A particular focus of this study was on users’ interactions with a chatbot in a public setting and in a private environment. Focusing on the event of users’ interaction with a chatbot interface, seventy-five interactions were videotaped in this study, in which fifteen users were asked to interact with the chatbot “Lucy” for their language learning. The video-stimulated post interaction interviews with participants provided complementary data for understanding their experiences with the language learning system. Analysis of twenty-five interactions selected from a total of seventy-five revealed five main factors of chatbot language tutor interface design and their relative significance in the process of users’ meaning making and knowledge construction. Findings showed that users’ sensory, emotional, cultural, linguistic and relational engagement influenced their responses to the chatbot interface, which in turn, shaped their learning processes. Building on a theoretical framework of cognitive emotions and emotions in learning, this study documented users’ language learning processes with the chatbot language learning interface by investigating users’ experiences. The findings and techniques resulting from this study will help designers and researchers achieve a better understanding of users’ experiences with technology and the role of emotions in the processes of learning when using technology and assist them to improve the design of language learning environments.

During the past few decades, there has been increasing attention to multimodal adaptive language learning interface design. The purpose of this study was to examine users’ experiences with a chatbot language learning interface through the lens of cognitive emotions and emotions in learning. A particular focus of this study was on users’ interactions with a chatbot in a public setting and in a private environment. Focusing on the event of users’ interaction with a chatbot interface, seventy-five interactions were videotaped in this study, in which fifteen users were asked to interact with the chatbot “Lucy” for their language learning. The video-stimulated post interaction interviews with participants provided complementary data for understanding their experiences with the language learning system. Analysis of twenty-five interactions selected from a total of seventy-five revealed five main factors of chatbot language tutor interface design and their relative significance in the process of users’ meaning making and knowledge construction. Findings showed that users’ sensory, emotional, cultural, linguistic and relational engagement influenced their responses to the chatbot interface, which in turn, shaped their learning processes. Building on a theoretical framework of cognitive emotions and emotions in learning, this study documented users’ language learning processes with the chatbot language learning interface by investigating users’ experiences. The findings and techniques resulting from this study will help designers and researchers achieve a better understanding of users’ experiences with technology and the role of emotions in the processes of learning when using technology and assist them to improve the design of language learning environments.

Chattbottar har den senaste tiden, genom snabbmeddelandetjänsternas stora tillväxt, blivitallt populärare. Chattbottar används idag till allt från automatisk kundtjänst tillnöjesinriktade ändamål. Numera finns flera olika ramverk för att skapa chattbottar med olikaegenskaper.Syftet med detta arbete var att undersöka hur chattbottar kan integreras i existerandesnabbmeddelandetjänster på ett underhållbart och flexibelt sätt. En prototyplösning förmeddelandetjänsten Briteback skapades med opensource-ramverken React, Node.js och Deepstream. Rapporten beskriver för- och nackdelar med den valda lösningen specielltavseende flexibilitet och underhåll.

During the past few decades, there has been increasing attention to multimodal adaptive language learning interface design. The purpose of this study was to examine users’ experiences with a chatbot language learning interface through the lens of cognitive emotions and emotions in learning. A particular focus of this study was on users’ interactions with a chatbot in a public setting and in a private environment. Focusing on the event of users’ interaction with a chatbot interface, seventy-five interactions were videotaped in this study, in which fifteen users were asked to interact with the chatbot “Lucy” for their language learning. The video-stimulated post interaction interviews with participants provided complementary data for understanding their experiences with the language learning system. Analysis of twenty-five interactions selected from a total of seventy-five revealed five main factors of chatbot language tutor interface design and their relative significance in the process of users’ meaning making and knowledge construction. Findings showed that users’ sensory, emotional, cultural, linguistic and relational engagement influenced their responses to the chatbot interface, which in turn, shaped their learning processes. Building on a theoretical framework of cognitive emotions and emotions in learning, this study documented users’ language learning processes with the chatbot language learning interface by investigating users’ experiences. The findings and techniques resulting from this study will help designers and researchers achieve a better understanding of users’ experiences with technology and the role of emotions in the processes of learning when using technology and assist them to improve the design of language learning environments. / Education, Faculty of / Curriculum and Pedagogy (EDCP), Department of / Graduate

Magíster en Ingeniería de Negocios con Tecnologías de Información / El presente documento tiene por finalidad la presentación del trabajo desarrollado en torno a una problemática de negocio presente en el Observatorio del Gasto Fiscal de Chile, la cual se aborda desde un proyecto de mejora de procesos como parte del MBE de la Universidad de Chile. El Observatorio del Gasto Fiscal es una iniciativa sin fines de lucro fundada en septiembre de 2015, la cual nace con la finalidad de generar mayores capacidades ciudadanas de observar, controlar y evaluar el buen uso de los recursos públicos. En este sentido, y entendiendo la amplia asimetría de conocimiento técnico sobre el presupuesto, el Observatorio ha desarrollado una plataforma web donde mantiene diversos análisis gráficos y herramientas interactivas que promueven la educación y el interés ciudadano sobre el gasto del Estado. Sin embargo, a pesar de su propuesta de valor innovadora, su desarrollo actual de productos no alcanza a cubrir de forma satisfactoria las necesidades específicas de información, sobre todo considerando a usuarios que requieren datos concretos y que no tienen la autonomía ni el tiempo necesario para encontrar las respuestas. En base a esta problemática y considerando que el Observatorio no cuenta con los recursos necesarios para destinar profesionales directos al servicio de los usuarios, este proyecto aborda el diseño de un servicio de atención automática, en base a un agente virtual de tipo chat capaz de establecer una conversación en lenguaje natural con el usuario respondiendo de forma autónoma sus preguntas. Como beneficios derivados de este servicio se establecen la reducción del tiempo y la certeza de obtener información correcta para usuarios actuales y potenciales. Para el desarrollo de este proyecto se aplicó la metodología de la Ingeniería de Negocios, sobre la cual, a partir de un análisis de la arquitectura empresarial de la organización, se definió una nueva línea de servicio dentro de su cadena de valor principal, resultando en el diseño y modelamiento de nuevos procesos relativos a la gestión de la relación de los usuarios, así como procesos encargados de la gestión, control y producción de la entrega de la información del sistema automático. Adicionalmente, este proyecto diseñó la lógica de reconocimiento de lenguaje natural de los usuarios subyacente al servicio, así como las reglas de negocio para la articulación coherente de respuestas sobre la estructura del presupuesto público. Finalmente, en relación a la evaluación económica del proyecto, cabe mencionar que a través de la implementación de un prototipo fue posible estimar una rentabilidad social positiva, reflejada por un VAN social de $56.442.968 y una TIR social de 86%.

En civilingenjör som utbildas idag förväntas ofta att lära sig programmering och för att klara en civilingenjörsutbildning krävs att man ska kunna programmera på många olika nivåer. I samband med den växande populariteten av chatbotar inom utbildning och bland företag så finns det skäl att undersöka hur en chatbot hade kunnat assistera studenter som läser programmeringskurser. Det stora antalet studenter och den begränsade tillgängligheten av lärarassistenter bidrar till lägre nivåer av personlig uppmärksamhet och support inom programmeringskurser. Studien syftar därför till att förstå om en meny-baserad chatbot kan vara ett effektivt komplement till laborationstillfällen. Först utvecklades en meny-baserad chatbot som webbapplikation med hjälp av React, Javascript, HTML5 och CSS3. Den meny-baserade chatbotens prestanda utvärderades sedan under ett simulerat laborationstillfälle där även deltagarnas (N = 6) upplevelser med chatboten samlades in. Resultatet visade att den meny-baserade chatboten användes totalt fyra gånger per student i snitt och kunde hjälpa studenterna att lösa deras problem i 62.5% av fallen, samt att den var generellt uppskattad av studenterna. Genom att möjliggöra hjälp för flera studenter samtidigt och att bättre förbereda studenterna inför deras möten med lärarassistenter uppvisade chatboten potential för att öka laborationstillfällets effektivitet, men att ytterligare undersökningar krävs för att säkerställa detta påstående. Vidare upptäcktes flera potentiella implementationsmöjligheter av meny-baserade chatbotar i denna miljö. Exempelvis kan olika visualiseringar av användardata som chatboten samlar in sammanställas för att få en inblick i studenternas lärande och vilka aspekter av undervisningen som är mer resurskrävande. Potentialen för användandet av meny-baserade chatbotar för programmering inom högre utbildning är därför tydlig. / An engineer educated today is often expected to learn programming, and to successfully complete an engineering education, one needs to be able to code at various levels. Given the growing popularity of chatbots within education and business contexts, it is worth investigating how a chatbot could assist students taking programming courses. Furthermore, the imbalance between the large number of students and the limited availability of teacher assistants leads to a low level of personalized attention and support for students in programming courses. The study therefore aims to understand if a menu-based chatbot can be an effective complement to lab sessions. Firstly, a menu-based chatbot was developed as a web application using React, Javascript, HTML5, and CSS3. The performance of the menu-based chatbot was then evaluated during a simulated lab session where data was collected as well as participants' (N = 6) experiences with the chatbot. The results showed that, on average, the menu-based chatbot was used four times per student and successfully helped students solve their problems in 62.5% of cases. The chatbot was also generally appreciated by the students. By enabling assistance for multiple students simultaneously and better preparing them for interactions with teaching assistants, the chatbot demonstrated potential for increasing the effectiveness of lab sessions. However, further investigations are required to substantiate this claim. Additionally, several potential implementation possibilities for menu-based chatbots in this environment were discovered. For example, different ways of visualizing user data collected by the chatbot could be compiled to gain insights into students' learning and identify more resource-intensive aspects of teaching. The potential for using menu-based chatbots in programming within higher education is therefore evident.

Yes / Older adults use their smartphones to learn new material but few studies examined their learning with the presence of a chatbot in a smartphone. We developed a three-week self-paced learning module on three topics (chatbot, QR scanner, Google Drive) using their smartphone. Our aims were to examine participants’ self-paced learning accuracy while exploring older adults acceptance of the chatbot. Twelve participants participated in this study (Mage: 64.75 years) for three weeks at their homes individually. Results showed that they had low accuracy for the chatbot but higher accuracy for the other two. Qualitative analyses indicated that participants disliked the chatbot and that good clarity in our instructional videos and slides may have contributed to the low acceptance for the chatbot. Our findings indicated that self-paced learning is feasible with slides and videos only, and to create more chatbot-related assessments to increase chatbot usage. / Newton Fund Institutional Links grant ID: 331745333.