Skip to Content
Last modified by Demi Breen on 2023/04/09 14:59
From version 6.1
edited by Liza Wensink
on 2023/04/04 15:37
Change comment: There is no comment for this version
To version 2.1
edited by Liza Wensink
on 2023/03/19 21:31
Change comment: There is no comment for this version

Summary

Details

Page properties
Title
... ... @@ -1,1 +1,1 @@
1 -7. Persuasiveness of conversational agents
1 +Persuasiveness of conversational agents
Content
... ... @@ -1,38 +1,28 @@
1 -Since the focus of our design is to motivate a PwD to follow along on a walk in the garden together with the robot, we will most likely need to take persuasiveness into account. Persuasiveness in human-human interactions consists of persuasion tactics and behaviors that might make a certain person more or less convincing. When it comes to human-robot interactions these aspects also come into play, with the added challenge of the agent not being able to employ all tactics a human might be able to do. Below we, therefore, dive into persuasiveness in conversational agents and what could be essential when designing a system with an objective like this.
1 +**Article: **Persuasive Conversational Agent with Persuasion Tactics. [[https:~~/~~/link.springer.com/chapter/10.1007/978-3-642-13226-1_4>>https://link.springer.com/chapter/10.1007/978-3-642-13226-1_4]]
2 2  
3 -Generally, for a conversational agent to be persuasive and influence a person's behavior it needs to be able to adapt to the outcomes of the conversation and the interactions it has with the human, as would a human who wants to be persuasive, according to Narita and Kitamura [1]. When it comes to designing the agent itself, several models can be used. The general approach is to select the response and the rule of replying that is most likely to lead to the desired goal [1].
3 +A number of studies had been done regarding the persuasiveness of conversational agents and how convincing an agent actually might be to a human person. This paper highlights that for a conversational agent to be persuasive and influence a person's behavior they need to be able to adapt to the outcomes of the conversation and the interactions it has with the human, as would a human who wants to be persuasive and convincing.
4 4  
5 -This could be done through a conversational model which can be represented as a state transition tree, using a goal-oriented approach. The different statements that can be given by the robot are then represented as links to change from one state to another [1]. Since these interactions imply a dialogue there would be two different types of states: human states and agent states which are interconnected in conversation paths. These paths represent the flow of conversations, beginning with an initial state and ending with either success or failure [1]. When the input from the human links to an agent state the agent chooses a statement that leads the agent's state to the human state with the greatest probability of success [1]. The model is updated when an input is provided that the agent is not familiar with. This causes the conversation path to branch and the model updates the probability scores [1]. While it is not feasible to develop a full-scale conversational model for the sake of our design for this project, this clearly illustrates the general approach to persuading with the help of a conversational agent. A clear goal is set for the interaction and the agent attempts to act accordingly, in steps that bring the conversation closer to the goal.
5 +This article also contains useful information when it comes to designing a persuasive conversational agent using the Wizard of Oz method. I won't describe it in detail here, but if it is needed there is some information to be taken from the article.
6 6  
7 -The persuasiveness-objective in the given study centered around showing the participant two different cameras, A and B. The purpose of the persuasion was to make the user change their initial choice [1]. The process of persuasion was done according to Table 1 below [1] where the general tactic was to try to convince the user to choose the other camera by explaining why the concerns they raise might not be relevant, such as explaining that either the pixels or the stabilizer do not carry much weight [1]. The model already has set predictions of what a user might inquire about and has pre-written responses that might change the opinion of the user [1]. In our case maybe we could also attempt to catch some reasons somebody might not want to go walking for example, and then try to explain why those reasons are not relevant or important to try to persuade the user to actually go out on the walk.
7 +When it comes to designing a persuasive conversational agent, there are several models that can be used. The general approach is to select the response and rule of replying that is most likely to lead to success.
8 8  
9 -[[image:attach:flowchart.PNG||height="702" width="416"]]
9 +**Goal-oriented conversational model. **
10 10  
11 -The study does, however, continue by mentioning that the Wizard of Oz approach, where the robot is simply controlled by a human in a wizard-like fashion, managed to persuade 25/60 users and the conversation agent based on the model only managed 1 out of 10 users [1]. A necessary takeaway here is to remember that designing a persuasive conversational agent consists of two important aspects, which will be crucial in the design of our project also. These are:
11 +(Quotes from the article)
12 +\\- The conversation model can be represented as a state transition tree where a statement is represented as a link to change a state from one to another.
13 +- Two different types of states, agent states and user states (the human). 
14 +- They are interleaved on a conversation path.
15 +- A conversation path represents the flow of conversation between the agent and one or more users and begins with the initial state and terminates with either success or failure.
16 +- If the input matches a statement on a link to an agent state, the agent chooses a statement that links the agent state to a user state with the greatest success probability.
12 12  
13 -* having the robot follow general human conversational rules
14 -* applying persuasiveness tactics [1].
18 +This might not be necessarily a structure we need to implement in its entirety, but some information could definitely be taken from it.
15 15  
20 +**Updating conversation model. **
16 16  
22 +When updating the above conversation model needs to be updated, it goes according to the following:
17 17  
18 -**Article: **Persuasive Conversational Agent with Persuasion Tactics. [[https:~~/~~/link.springer.com/chapter/10.1007/978-3-642-13226-1_4>>https://link.springer.com/chapter/10.1007/978-3-642-13226-1_4]]
24 +- When input from the user does not match any statement on the stored conversation path, the conversation path is branched and the success probability scores are updated depending on persuasion success/failure. (Once again, maybe not something we will be able to implement but can try to somehow mimic the idea of).
19 19  
26 +The article does, however, continue by mentioning that the Wizard approach where the robot is simply controlled by a human in a wizard-like fashion managed to persuade 25/60 users and the conversation agent based on the model only managed 1 out of 10 users. It is necessary to remember that designing a persuasive conversational agent consists out of two important aspects - having the robot follow genera human conversational rules, but also applying persuasiveness tactics. I will attempt to clarify these tactics a bit below.
20 20  
21 -**Article**: Applying Psychology of Persuasion to Conversational Agents through Reinforcement Learning: an Exploratory Study. 
22 -[[https:~~/~~/ceur-ws.org/Vol-2481/paper27.pdf>>https://ceur-ws.org/Vol-2481/paper27.pdf]]
23 -
24 -This study concerns itself with agents trying to induce a healthier diet into the human they are attempting to persuade, which could be somewhat similar to what we are attempting to do.
25 -
26 -This study mentions: 
27 -"Three relevant psychosocial antecedents of behaviour change are the following: Self-Efficacy (the individual perception of being able to eat healthy), Attitude (the individual evaluation of the pros and cons) and Intention Change (the individual willingness of adhering to a healthy diet). These psychosocial dimensions cannot be directly observed and need to be measured as latent variables. To this purpose, questionnaires are used..."
28 -
29 -What was later done during the test was...
30 -
31 -"In a subsequent phase (i.e. message intervention), participants were randomly assigned to one of four groups, each receiving a different type of persuasive message: gain (i.e. positive behavior leads to positive outcomes), non-gain (negative behavior prevents positive outcomes), loss (negative behavior leads to negative outcomes) and non-loss (positive behavior prevents negative outcomes)." Could be something that can be considered during the persuasion stage.
32 -
33 -All this together is maybe be a bit much for us to implement. These questionnaires are quite lengthy and complicated to design and evaluate, since these aspects need to be monitored through latent variables. While we shouldn't and can't implement this in our project currently, it might be good to include as a side point when it comes to designing the complete system.
34 -
35 -Further this article mostly descends into how to translate this different aspects and variables into a Bayesian network and then training the agents using RL, which is not relevant for this course even if it is interesting. Once again, could maybe be mentioned as a side note.
36 -
37 -
38 38  
flowchart.PNG
Author
... ... @@ -1,1 +1,0 @@
1 -XWiki.lwensink
Size
... ... @@ -1,1 +1,0 @@
1 -63.3 KB
Content