STORYLINE AND DRAMA Mike Chu, Joey Blekicki, Stephen
STORYLINE AND DRAMA Mike Chu, Joey Blekicki, Stephen Kish STORYLINE AND DRAMA History of Narrative Development Dialogue Management Gameplay and Story New Research in Storyline Development HISTORY OF NARRATIVE DEVELOPMENT PROPP AND THE FORMALIST
Morphology of the folktale Discovered stable structures in Russian folktales Transgression, deception, struggle, punishment, wedding, etc. A sequence of 31 narrative functions PROPPS APPROACH Narrative functions are invariant elements and are independent from the characters that execute them There are a limited number of narrative functions
Functions always occur in the same order No backtracking allowed LIMITS TO PROBBS MODEL No branching to allow for alternate stories It is not interactive GREIMAS
Introduced the role-based analysis of narratives Define characters not for who they are, but for what they do Defined roles: subject vs. object Also identifies semantic fields: love, greed, etc BARTHES AND THE INTERPRETATIVE CODES Defined scenes in terms of action ramifications Actions have the dimension of semantic field Not restrained to a specific occurrence
BATHES CONT. Some actions do not require a specific sequence The result of a murder and the reason for the murder Theory based on five codes:
ACT (ACTion) REF (REFerence) SYM (SYMbolic) SEM (SEMantic) HER (HERmaneutic) BREMOND His theory centered around the concept of character roles Opposition
between Agent and Patient BREMOND CONT. Characters can alternate between the agent and patient role A patient can be prompted into taking action A narrative process affects a patient by: Influencing
their awareness of a situation Altering the situation (improving or worsening) Agents Voluntary-purposefully initiates a goal-oriented process Involuntary-narrative impact derives from unintended side-effects of actions BREMOND CONT. AGAIN
Also introduces character psychology Characters The Matrix have beliefs, motivations, and goals MORE Patient influencing process
Their actions can influence the outcome of the agent Portal DIALOGUE MANAGEMENT DIALOGUE MANAGERS History and beginnings Colossal
Cave Adventure Progressions Neverwinter Nights Dialogue Manager Center
of interactive language systems Speech or text based Responsible for what is said and is talked about Based on historical information, goals, possible actions DIALOGUE MANAGER (DM) Finite state machines, frames, stacks, inference engines, planners Several definitions historically
Process decide what is said at time steps Similar to goal oriented control structures DM favored system initiative vs. user initiative vs. mixed initiative System: system drives dialogue
Final Fantasy Tactics User: user drives dialogue, system responds Mixed: user chimes in, system responds and redirects FINITE STATE MACHINES Most popular Dialogue Management Paradigm and implementation technique of commercially spoken dialogue systems Neverwinter Nights
FINITE STATE MACHINES Distinctive advantage with spoken language Use specially tuned acoustic language model for each dialogue state Know what utterances to expect Makes auto speech recognition task easier
Suited for special situations System has dialogue initiative, dialogue states, dependencies between are defined and not many (FSM) EXAMPLE CONT. Milk Br Swe et SA Tea
Choo se Food Choo se Coffe e No Milk Br Br Swe
Swe et et Milk Br Latt Latte e Sugar South African
Cak e Br Brazilian No Sugar Italian It Br Br Swe Swe
et et Latt Latte e Br Not Swe et No Milk Br Br
Br Pur Pure e FINITE STATE MACHINES Provides straightforward way from task breakdown to DM implementation Easy to check uncovered conditions, shortest paths, cycles etc Difficult lies in growth Problematic
with interrupts to system directed dialogue Standard: ignore -> steer user back to original (FSM) SUMMARY Good for simple, informative characters Implementing task oriented subdialogues Familiar to game developers Good starting point when implementing characters with dialogue capabilities Limitations Need
all data to adhere to ordering constraints Any new information not expected is discarded Examples: Text based FRAMES Second most popular dialogue modeling technique Foundational paradigm of VoiceXML (Expand) Widely used in commercial dialogue systems Used to fill a form and populate and/or query a database Typical applications
Transport timetable info, call routing (expand) FRAME BASED DIALOGUE SYSTEM (FBDS) Gets its name from way info is gathered from user Frame is viewed as object in object oriented paradigm with no defined methods Frame keeps track of wanted info Algorithm determines what to do to fill missing items
Prompts initial question, fill as many slots as possible with users current utterances, ask questions to clarify and fill remaining slots FBDS Necessary to keep track of info and create clarification questions Need
to keep confirmation flag Need confidence values associated with slots Similar to learning tables Stronger in case of spoken dialogue systems Need to address ambiguity and interpretation problems FBDS Allow more efficient and natural interaction
System able to use info not explicitly asked for but relevant to the frame Ease burden to software engineers Allowed to specify dialogue rules for each frame Management algorithm generates appropriate dialogue moves dynamically Limitations To
use automatic speech recognition component must be robust Needs to be able to deal with utterances used to describe everything in any given frame Goes for natural language understanding module and embedded clarifications Unable to deal with info that falls out of current frame but is still relevant and supported Not used in video games more so commercial products Human
and computer interactions STACKS Provide natural way to change topic of conversation and resume halted one Any new conversation is pushed over the old and then the old is popped once the new is done Can be integrated or independent COMIC system
Uses stack and augmented FSM as basis of DM AFSMs called Dialogue Action Forms (DAFs) Has ability to execute arbitrary action in state transitions Wait for arbitrary and external info Indexing terms like keywords STACKS DM changes topics through DAF creation, indexing and selections DAFs created with properties such as verbs,
nouns, entities and restrictions on world properties Combo of keys are put in index When system receives utterance makes key from bits of info from sentence Then selects DAF most closely matched and is put on top of current one When finished it is popped and pervious one resumes STACKS Limitations
Auto speech recognition needs to have one general layer capable of identifying all utterances that lead to a topic or task shift Natural language understanding module needs to be able to spot keywords, dependencies, entities that signal topic shifts Tuning the indexing and retrieval mechanisms is challenging task in itself Needs more sophisticated language generation module capable of summarizing what was said before and introducing appropriate cues and intros to resume previous converstations
STACKS Implementation Most games use this type of dialogue manager Tutorials Sequences for more information Ex) Dragon Age, Brave Fencer Musashi INFERENCE BASED SYSTEMS
Has four components Knowledge base is composed of declarative rules in logical formulism Knowledge base, inference engine, working memory, facts selector
Propositional logic, first order logic Inference engine responsible for finding valid proof of a fact Supports unification, forward/backward chaining Working memory is where facts of current interests are kept Facts selector is an algorithm that chooses and combines facts of interest before put in the inference system NICE game system hybrid system
IBDS Limitations Difficult to use tuned auto speech recognition model for different dialogue parts with an IBDS Language understanding module needs to provide enough information to populate working memory with relevant facts Having good knowledge base information to guide interpretation of utterance
Mass Effect Blue: Charm, Red: Intimidate PLAN BASED SYSTEMS Integral part of research and cutting edge commercial dialogue management systems Basic structure Set
of operators and procedures to find sequence of operations that achieve one or more goals Operations usually specified in terms of preconditions and effects Two basic common uses Encoding speech acts/DM output directly to operators actions To select facts of interest to be fed into the system PLANNED BASED SYSTEMS
Ordering exists between actions to organize; need to know how many people and where will they sit Offers same complications for auto speech recognition and natural language understanding as an interface based system TRIPS Final Fantasy 9: Cooking (1:39) GAMEPLAY AND STORY Brief history of story in games Modeling faction Interactions
Applications to RPGs Faction modeling in action STORY IN GAMES More of a recent development Earlier games relied heavily on gameplay (Megaman, Pong, Super Mario Brothers) Still plenty of games with limited base story: Super Mario Galaxy, Katamari, Call of Duty, Sports Games, Fighting Games. MODERN GAMES Games are now expected to have at least
some type of story. RPG storylines have been growing in scale, and just keep getting larger. MODEL PARAMETERS Assuming two factions (X,Y), each faction gets a parameter (x,y) where x or y is that factions level of cooperation towards the other. Higher parameter value means more cooperative. Factions may also have other
parameters for belligerence or pacifism. All the parameters are evaluated to decide the factions behavior towards EQUATIONS TO USE PARAMETERS (BASED ON MODELING AN ARMS RACE) Equations describing intended behavior: x = ky ax + g y = lx by + h k and l are fear constants (mutual)
a and b are restraint constants g and h are grievance terms If ab > kl, there is equilibrium. If ab < kl there is unstable equilibrium. REINTERPRETED FOR RPGS Same basic equations X = Ky Ax + G Y = Lx By + H K and L are belligerence factors A and B are pacifism factors G and H are friendliness towards the other faction
If the result is above equilibrium, they are in cooperation, if they are below, they are in competition, and on or around eqilibrium is neutrality. BEHAVIOR IN ACTION Parameters may be applied toward random encounters with other factions. More hostility = more difficult battles. Allied factions assist in battles
If factions are cooperative, maybe a negotiation encounter will occur. Also effect conversations as shown in dialog trees. EXAMPLES IN MODERN GAMES Fable Fallout (Good/Evil) Neverwinter Nights Series Oblivion Almost every NPC belongs to some faction: Fighters guild, Mages Guild, Thiefs Guild, Dark Brotherhood, Arena.
Also minor guilds: Blackwood Company, Knights of the Nine, The Blades, Order of the Dragon, etc. The factions standing towards the player effects dialogue and aggressiveness. ENDINGS OF GAMES Some games use character or faction relations to change the endings(Star Ocean Series, Heavy Rain). Evaluate relationship values and present different scenarios based on those values, allowing different endings for multiple playthroughs.
STORY IN GAMEPLAY Normally, player expects story to effect gameplay, newer games make it work so that gameplay also effects the story. Example: http://www.youtube.com/watch?v=SQCBLsJhc Do#t=02m35s RESEARCH IN STORYLINE DEVELOPMENT INTERACTIVE STORYTELLING
RESEARCH Experimenting how interfering with actions can affect outcomes of a storyline STORYLINE BASED ON AWARENESS AND FEELING NARRATIVE GENERATION THROUGH POV
period or shareholding unlike other international business companies like UK or Dutch. UNILATERAL CREDIT RELIEF FOR FOREIGN TAXES, NO CFC LEGISLATION, NO THIN CAPITALISATION RULES OR TRANSFER PRICING RULES. NO WHT ON DIVIDENDS PAID . EVEN IF THEY ARE PAID...
Induction « Opéron Lactose » Synthèse des enzymes nécessaires au métabolisme du lactose Chez les procaryotes Répression « Opéron Tryptophane » Synthèse des enzymes impliquées dans la biosynthèse du tryptophane Structure de l'opéron « lac » L'opéron lactose d'Escherichia coli...
We accomplish our vision through grantmaking, policy advocacy, and strategic communication. Reduce health disparities related to social and economic factors (health equity) Social Determinants of Health. Engage Kansans to improve the health of our state (civic health) Actionable Data. Media...
History of Continental Drift. John Tuzo Wilson. Canadian geophysicist. mapped transform boundaries along the mid-ocean ridge in the Atlantic. Suggested that the oceanic crust is doing more than growing but also sliding past each other. Coining and using the phrase...
If have a curved surface (wave front) and the bottom of the surface moves slower than the top the surface will have a torque on it which will tend to turn the curved surface towards the slow speed area. ......
uTor must negotiate to use 2 byte cell ID. Compatible with existing Tor deployments. Preliminary Evaluation of uTor. Tor configuration. 3 Relays. 3 Directory Authorities. A single proxy. OP takes the same path through the 3 relays. Version .2.5.-alpha-dev (git-7c670895b02ba731)
Writing Effective Success Stories Kimberly Keller, Ph.D., CFLE ... Objectives Define a Success Story, and the rationale for using it Understand different types of Success Stories Identify key audiences with whom to share Success Stories Develop strategies for communicating Success...
May 2008 - report by Dr. Linda Aikens & colleagues confirmed findings from 2003 study showing for every 10% increase in proportion of BSNs was associated with a 4% decrease in the risk of death Studies have also demonstrated in...
Ready to download the document? Go ahead and hit continue!