Conversational IIS Flashcards Preview

Interactive Information Systems > Conversational IIS > Flashcards

Flashcards in Conversational IIS Deck (24):
1

Definition Agents

Agents are autonomous, active computer processes that possess some ability to communicate with people and/or other agents and to adapt their behavior (Benyon, 2014).

2

Agents - ‘Strong view’ of Artificial Intelligence area

- Agents have beliefs, desires, and intentions
- Agents can plan, learn, adapt and communicate

3

Agents - ‘Weak view’ of Human-Computer Interaction area

Agents serve as ‘indirect managers’ --> interface agents

4

Conversational Agents comprise two main parts

Agent: Agents are autonomous, active computer processes that possess some ability to communicate with people and/or other agents and to adapt their behavior.

Converstational Interface: Conversational interfaces are interfaces that enable people to interact with smart devices using conversational spoken language.
• Voice (e.g. personal assistants)
• Text (e.g. chatbots)

5

Definition - Conversational Agent

Conversational agents, or chatbots, provide a natural language interface to their users. (Kerly, Hall, & Bull, 2007, p. 177)

6

Definition - Chatbot

Chatbot is defined as a program that emulates human conversation and enables natural-language conversations with computers. (Lu, Chiou, Day, Ong, & Hsu, 2006, p. 576)

7

classification of conversational agents

Classification of conversational agents based on two dimensions (Gnewuch et al., 2017)
- Mode of communication: text vs. speech
- Context: general-purpose vs. domain-specific
siehe Folie 10

8

General-Purpose, Text-Based Conversational Agents

e.g. Cleverbot
- Interaction via text messages (“chatting”)
- Conversational agents can basically converse
about any topic
- Developed to win the Loebner prize / to pass the Turing test

9

Domain-Specific, Text-Based Conversational Agents

 Interaction via text messages (“chatting”)
 Conversational agents are trained to converse
about specific topics related to their domain
 Used for entertainment or in customer service, healthcare, banking, ...
 Can be found on websites and on many messenger platforms such as Facebook messenger, Telegram, Slack, ...

10

General-Purpose, Speech-Based Conversational Agents

 Interaction via speech
 Can basically answer any question and support users in finding information or accomplishing basic tasks
 Found on almost all new mobile devices
 Often called virtual or digital assistants

11

Domain-Specific, Speech-Based Conversational Agents

 Interaction via speech
 Deployed to assist in a specific situation
 Examples can be found in modern cars or in border screening processes

12

Classification of Conversational Agents (cont’d)

Natural language interface
- Text (e.g., Facebook bot)
- Speech (e.g., Amazon’s Alexa)
- Text & Speech (e.g., Microsoft Cortana)
Visual interface
- Disembodied (e.g. Apple’s Siri)
- Embodied (e.g., Facebook bot)
Dialog system
- Pattern matching (i.e., simple chatbots)
- Deep understanding (e.g. Apple’s Siri)

13

Visual Representations of Conversational Agents

 Conversational agents often have some kind of visual (human-like) representation
 These agents are called embodied conversational agents
 Embodied conversational agents “are human-like virtual characters that can provide direct
and interactive communication with humans”.

(Garrido, Barrachina, Martinez, & Seron, 2017, p. 2)

14

Technical Components of Conversational Agents

Main components
 Speech recognition: converts the speech input
into a string of words
 Spoken language understanding: interpret the user’s input and to extract a representation of its meaning
 Dialog management: tracks the state and flow of the conversation and controls how the system responds to the user’s input
 Response generation: Formulate a response in natural language
 Text-to-speech synthesis: converts the response string into speech output

15

What is Driving the Evolution of Conversational Agents?

 Technological advances in Artificial Intelligence, Machine Learning, Natural Language Processing ...
 Proliferation of messaging apps and available chatbots on this platforms, e.g., over 100,000 chatbots have been created in less than one year on Facebook Messenger (Johnson 2017)
 New device technologies are available at a reasonable price (e.g., smart phones, smart devices)
 Increased connectivity (e.g. free WiFi, 4G, etcs)
 Huge investments by major technology companies
 By 2020, the average person will have more conversations with bots than with their spouse (Gartner
2016, p. 3)

16

Definition - personal digital assistant

A personal digital assistant is a meta-layer of intelligence that
sits on top of other services and applications and performs actions
using these services and applications to fulfill the user’s intent.

17

personal digital assistant - User’s intent could be

 explicit, where the user commands the
system to perform an action, or
 inferred, where the agent notifies or makes suggestions upon evaluation of one or more triggering conditions it has been tracking.

18

personal digital assistant - PDAs make use of some core set of technologies, such as

 machine learning,
 question answering, dialog management
 speech recognition, language generation, text-to-speech synthesis
 data mining, analytics, inference, and
personalization.

19

PDAs support / assist users

 getting things done (e.g., setting up an alarm/reminder/meeting, taking notes, creating lists)
 provide easy access to personal/external structured data, web services, and applications (e.g., finding the user’s documents, locating a place, making reservations, playing music).
 In their daily schedule and routine by serving notifications and alerts based on contextual information (e.g. time, user location, and feeds/information)
 Make users more productive in managing his or her work and personal life.

20

Computer Are Social Actors (CASA)

 Users mindlessly apply social rules and expectations in their interaction with computers that use natural language or display other human characteristics (Nass et al. 1994, Nass & Moon 2000)
 Reeves and Nass introduced the CASA-paradigm and showed that human-computer interaction is fundamentally social and natural (Reeves & Nass, 1998). They analyzed this phenomenon in various studies and proved that humans react to computers in a similar way as they do towards other humans (Nass et al., 1994; Nass & Moon, 2000; Reeves & Nass, 1998).

21

Social Cues

 A cue is “any feature of the world, animate or inanimate, that can be used by individuals as a guide to future actions”
 Help to clarify people’s meanings and intentions
 Influence various social processes (e.g., communication)
 Examples in human-human interaction:
 Facial expressions (e.g., smiling)
 German forms of address (e.g., Du vs. Sie)
 ...

22

Definition - Social Cues

Social cues as design features of a conversational agent that
automatically trigger emotional, cognitive, or behavioral reactions
by the user that are appropriate when directed at other humans
but inappropriate when directed at conversational agents

23

CASA paradigm

Social cues from computers trigger mindless responses from humans, no matter how rudimentary those cues are
- Cues: A technology possesses characteristics normally associated with humans. They trigger....
- Mindless Behavior: Application of various scripts, labels, and expectations in accordance with prior experiences. Result in ...
- Social Responses: Emotional, cognitive, or behavioral reactions similar to reactions shown during interactions with humans

24

Anthropomorphism

Anthropomorphism is the attribution of human-like physical or non-physical features, behavior, emotions, characteristics and attributes to a non-human agent or to an inanimate object