Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Chatbotsin Education
Vantaa, May 16th, 2019
• Innovation Unit Coordinator and lecturer at
Madrid Open University (UDIMA).
• My research and teaching focus on
the intersections of digital media and
society, with a special focus on innovation
and emerging technologies (mobile,
adaptative learning, wearable, smart
technologies...).
• I participate in national and international
projects, such as: AGE 50+, SOLA, EPBE,
CVTube, RefuSkills, GoMobile, INTRO 4.0,
B-SAFE.
Silvia Prieto Preboste
UDIMA, Spain
INFORMATION:
• 25,000 students per year
• 6,800 people in the CEF.- /
UDIMA Alumni Association
• More than 3,500 job offers
from the Job Bank
• 800 teachers and collaboratos
UDIMA is a CEF.- Business School project: The business school and the State Examinations center
have more than 40 years of experience
UDIMA presentation and academic offer
Marak Digital Marketing is a digital
advertising agency that specializes in
online promotion, marketing,
and emerging technologies.
They develop and implement
comprehensive online marketing
strategies to help their clients
to connect with the target audience.
Website: http://www.marak.es/en/
Marak Digital Marketing
Chatbots are computer programs that are able to interact
with people using language-based interfaces.
Searching for information, processing it and adapting it to
the user’s needs; answering an email following voice orders;
making a booking at a restaurant, or simply holding a
conversation are some examples of the types of interaction
that can occur between a person and a chatbot.
What is a Chatbot?
History of Chatbots
1972Parry
2001SmarterChild
2015Alexa
2017Woebot
1966Eliza
1981Jabberwacky
2013
Mobile CoachChatbots!
2016
Bots forMessenger
2010SIRI
1991Dr. Sbaitso
1950The Turing Test
1995A.L.I.C.E
2006IBM Watson
Conversational artificial intelligence,
the basic source of chatbots, thanks to which all management and natural
language processing (NLP) occurs. The first chatbots focused on the
interpretation and recognition of patterns and rules. The more advanced
chatbots implement deep learning processes to analyse the human input,
learn from conversations and generate as suitable a response as possible.
User interface (UI),
whereby the user can see or
hear the conversations with the
chatbot.
Conversational design,
which allows an artificial
interaction to be equipped with
human logic.
Elements of a chatbot
User experience (UX),
which allows a natural, intelligent
and coherent conversation to be
established.
(Nieves, 2018)
Types of chatbot
The interaction occurs through text inputs and outputs or voice inputs and outputs. With natural
language processing, written text can become oral and vice versa, which opens up communicative
possibilities in the interaction between person and chatbot (Clark, 2018).
E M BO DIE D CO N VE RS AT IONAL AGE N TS
The interface is represented by the shape of a body, or a face in the form of an avatar, which
interacts with the user and which may contain audio, text and other audiovisual and multimedia
representation resources (Allison, 2011).
PHYS I C A LThere is a third type of chatbot that is in the form of a physical robot, be it humanoid or otherwise.
B A S E D I N C HA T T E RB OX ES
Chatbot development platforms
IBM offers the Watson conversation service for language processing, which provides the creation of chatbots and virtual agents. It also has specific services for developing a number of conversation functions, such as speech to text and text to speech.
Microsoft has developed the Azure cloud services platform, which includes different AI and NLP tools to aidchatbot development. It also offers natural language services (language understanding intelligent service orLUIS) that process and analyse language (Yan et al., 2016).
Google offers TensorFlow, open software that provides materials and resources for developers. Google also has Cloud AI, a system for using or creating deep learning models. Finally, we have the Cloud Natural Languageapplication programming interface, focused on natural language processing and comprehension.
Amazon offers tools such as Lex, which enables developers to integrate chatbots in other mobile apps. Alexa is a cloud-based voice service especially designed for the Amazon Echo device.
Facebook is one of the main platforms where chatbots can work (Messenger and WhatsApp), buit it also offerswit.ai, desisgned for developers to create chatbots.
Other technologies for developers that enable the creation of chatbots are: Motion, Smooch, Gupshup, Botkit, Rasa (Futurizable, 2017), api.ai, Semantic Machines, Digital Genius, Chatfuel, Pypestream, Pandorabots, AgenBot, ChetterBot and ChatSript (Davydova, 2017).
Source: García et al. (2018)
Instant messaging apps that integrate chatbots
Messenger Facebook, Telegram, Slack, Skype, Twitter, Kik,
WeChat, Line, Viber, Snapchat and WhatsApp.
Main operational
virtual agents
Platforms
Source: García et al. (2018)
Siri (Apple)
Bixby (Samsung)
Google Assistant
(Google)
Cortana (Microsoft)
Types of chatbots in education I
Without educational intentionality:
these are chatbots that are
incorporated into teaching tasks of
an administrative nature (student
guidance and personal assistance)
and of a support nature (to answer
FAQs).
Depending on their nature
With educational intentionality:
these are designed to foster teaching
and learning directly.
Types of chatbots in education II
Administrative and management
tasks to foster personal productivity
Taking car of FAQs
Student mentoring
Depending on the functions carried out by chatbots:
Motivation
Practice of specific skills
and habilities
Simulations
Reflection and metacognitive
strategies
Student learning assessment
Types of chatbots in education III
Along the learning journey
Mentor
bots
Wellbeing
bots
Onboarding
bots
Learner
support bots
Invisible
LMS bots
Assessment
botsPractice
bots
Learner
engagement bots
Learner
engagement
Content
creationLearner
support
Content
curation
Adaptative
learning
Assessment
Wellbeing
Source: Clark, 2018
Examples of chatbots in educationAccording to intentionality
Without educational intentionality With educational intentionality
Academic guidance
and personal agent
Support
(FAQs)Tutors-support
Exercise
and practice
According
to tasks
Genie, Hubert, Ivy,
Pounce, CourseQ,
Differ, MOOCBuddy
Administrative and ma-
nagement tasks to foster
personal productivity
Genie, Ivy, Pounce,
Otto
Genie
Answering administrative
queries (FAQs)Genie, Ivy, Pounce,
CourseQ, Bot CEU
Cardenal, Herrera
Genie, Pounce, Jill
Watson, Bot CEU
Cardenal, Herrera
Mentoring Jill Watson Pounce, Jill Watson,
Differ, Ani, Botter
Duolingo, Pepper,
NAO
DuolingoConversations for
language learning
Motivation Differ, Ani, Botter Duolingo
Reflection and metacog-
nitive strategies
The Guardian of History
Replika
Student learning
assessmentThe Guardian of History
Replika
Pepper NAO
So
urc
e: G
arc
ía e
t a
l. (
20
18
)
Questions that should be
taken into account:
Considerations regarding designand configuration
Level of
'wisdom‘
of the
chatbot
Degree of
proactivity
Level of
'humanness' of
the chatbot and
its relationship
with the
student
Degree of
anthropomorphism
that the teaching
agent
61%
39%
Female Male
Master's Degree
in Educational Technology
Subject: Innovations in e-learning
(6 ECTS)
83 students
Research sample
PHASE 1
Design and programming
of the chatbot
PHASE 3
Peer assessment of the
MOOC’s proposals
PHASE 5
Assessment
by the teacher
PHASE 2
Design of
the MOOC
PHASE 4
Upload files to the
Virtual Classroom
SURVEY
Start of the
learning activity
with the students
Development of the learning activity
Big Data Analytics SEO/SE
Welcome to Marak Digital Marketing
Now let me show you some
of our services.
1. Question script design
We design the script of
questions, the transitions
between them and the types of
answers allowed.
Some examples of questions are:
• What is the title of your
MOOC?
• What is the MOOC about and
how would you describe it to
promote it?
The system also offers
information and clues to answer
the questions.
2. Peer assessment design
MoocBot is programmed to be
an assistant in the design of a
MOOC and also to assist in the
evaluation of the proposal of
another student’s MOOC.
The assessment assistant
includes questions such as:
• Are the different elements of
the MOOC exposed in a
clear and concise way?
• Do you think there is
coherence between the
different elements of the
proposal?
3. Programming the chatbot
The chatbot, called MoocBot, is
programmed and integrated into
the social network Facebook.
Test and check:
• Technical aspects.
• Transition between questions.
• Script.
The students can restart the
chatbot writing "Restart MOOC“
or "Restart assessment“.
Design and programming of the chatbot
The students accessed the MoocBot again to assess the
proposal made by another of their classmates.
MoocBot asked several questions to each student, such as:
• Does the MOOC include all the mandatory elements?
• Are the elements exposed in a clear and concise way?
• Is there coherence between the elements?
• What is the level of creativity and innovation of the
MOOC?
• Rating (1 to 10).
Peer assessment of the MOOC’s proposals
MOOC proposals were assessed by the teacher
according to the following criteria:
• Full MOOC design.
• Creativity in the selected topic and use of
multimedia elements.
• Coherence between the different elements of
the MOOC proposal.
• Clarity, coherence, originality, and innovation
in ideas.
• Applicability and viability of the proposal.
• Assessment of another classmate proposal.Assessment
of another
MOOC
proposal
MOOC
designed
Upload files to the Virtual Classroomand assessment by the teacher
The chatbot is a
good tool to carry
out a peer to peer
evaluation
The chatbot is a
good tool to
design a MOOC
Using the
MoocBot has
really helped you
to design the
MOOC
Totally agree
Agree
Neither agree nor disagree
Disagree
Totally disagree
Survey results
Totally agree
Agree
Neither agree no
disagree
Disagree
Totally disagree
The
interaction
with the
chatbot has
been easy for
you
The chatbot is
very useful in
the
educational
scope
The technical
level of
difficulty of
using a
chatbot is very
high
Survey results
What is your
level of
motivation
using the
chatbot?
What is your
overall rating
of the activity?
Very positive
Positive
Neither positive nor
negative
Negative
Very negative
Lot of motivation
Quite motivation
Neither motivation nor
unmotivation
Quite unmotivation
Very unmotivation
Survey results
Workshop
1 Before you build
your bot
2 Setup Welcome
Message
4 Creating
blocks
5
3 Setup Default
Replay 6
Elements
Setup
Keywords