Being able to narrow down a focus for my dissertation has been quite difficult as what I want to focus on isn't very easy to pinpoint. In the first worksheet I vaguely go over issues I'm interested in trying to make it sound like I have an idea and looking back its clear I really don't. However i was able to formulate my thought in an email to my 'possible' supervisor which I found helped me more than it probably help him to help me.
" My blog entry for worksheet one was rather difficult to produce as I thought I had a well thought out idea but as I wrote it it fell apart. The idea I want to investigate is creating an NPC that has a greater influence on a game's story, for example an NPC being able to confront a player on an action that the NPC disagrees with. The topics I have looked into are, NPC personality, Emotional components and different decision making processes but am not sure how I would narrow this down to a question or even if I could take aspects from each but I assume that would be to broad an area to study for this dissertation.
Practically I was looking to create a short text based story with the player making different decisions and the NPCs reacting to the decisions and attitude of the player in a way the would influence the story however how to go about making this a test I could draw a conclusion from has to this point eluded me."[07 October 2010]
Having had time to contemplate this and do some research I feel I have been able to identify the area I wish to study and although it is a broad area the focus should be able to be highlighted. Consequences. In many games at the moment there are no consequences for a players action, the game is just a linear story broken up between killing and cut scenes, trying to give emotional depth though uncontrollable movies and creating emotion by limiting it to anger and violence during game play. There are game out there that show a higher degree of choice and consequences but on a larger scale than I plan to look into, games like Fallout 3 and its newly released sequel Fallout: New Vegas allow you to take action where there are consequences that effect you gaming experience. However these decisions don't effect the players companions on an emotional level. It does effect the companion but in basic ways such as if you are an 'evil' character a good companion will leave your side and vise versa, also if you side with a faction that is an enemy to your companion they may leave your side. This is fine to help the story continue however it causes them to be a bit devoid of personality and emotion.
"Can an NPC's* personality and emotions be used to actively effect consequences in a game and possible the main storyline itself"
* NPC - Non-player Character.
There are many games (mainly RPGs) attempting to add a more diverse experience with the player's companion characters such as Bioware's Dragon Age Origins and Mass Effect(1 and 2) and Bethesda and Obsidian's Fallout 3 and Fallout: New Vegas. The area is also rich in AI literature as this uses personality programming, emotional components, decision-making and AI learning.
Practically implementing this would be to take various NPC's with different personalities and emotional components and put them into situations with the player with which the players decisions effect the NPC's 'views' and have the NPC react to the decision and implement a believable consequence for the decision.
To do this a number of NPC personality would have to be created then have each personality linked to an appropriate emotional state and possibly a background for each NPC. From there a simple text based story would be implemented with which the player would play through scenarios and the story will be determined by the NPC's personality. As a result some sort of learning algorithm will need to be implemented to determine exactly how the story is effected by the NPC's personality and emotions to allow for emergent behavior. Methods I could use are Fuzzy Logic, Genetic Algorithm or I could also use Support Vector Machines although the use of these methods will have to be evaluated and used at the focus of the dissertation rather than focusing on the techniques themselves.
By looking into what game are doing now in terms of the research question(RQ) and how other programmers have taken to the programming involved we can establish a baseline for what we expect to happen and define the result of the RQ by the emergent behavior that arises. The scope of this dissertation has a lot of different techniques used for a single purpose, I am hoping to use standard, credited methods for each to limit the variables and to give a more concise experiment. I am also hoping that the focus of this project is not skewed by the different factors that are involved. For consistency I should use a combination of all available techniques but that would be far beyond the scope of this project.
