Language Understanding for Text-Based Games Using Deep Reinforcement Learning

Total Page:16

File Type:pdf, Size:1020Kb

Language Understanding for Text-Based Games Using Deep Reinforcement Learning Language Understanding for Text-based Games using Deep Reinforcement Learning Karthik Narasimhan∗ Tejas Kulkarni∗ Regina Barzilay CSAIL, MIT CSAIL, BCS, MIT CSAIL, MIT [email protected] [email protected] [email protected] Abstract State 1: The old bridge You are standing very close to the bridge’s In this paper, we consider the task of learn- eastern foundation. If you go east you will ing control policies for text-based games. be back on solid ground ... The bridge In these games, all interactions in the vir- sways in the wind. tual world are through text and the un- derlying state is not observed. The re- Command: Go east sulting language barrier makes such envi- ronments challenging for automatic game State 2: Ruined gatehouse players. We employ a deep reinforcement The old gatehouse is near collapse. Part of learning framework to jointly learn state its northern wall has already fallen down ... representations and action policies using East of the gatehouse leads out to a small game rewards as feedback. This frame- open area surrounded by the remains of the work enables us to map text descriptions castle. There is also a standing archway of- into vector representations that capture the fering passage to a path along the old south- semantics of the game states. We eval- ern inner wall. uate our approach on two game worlds, Exits: Standing archway, castle corner, comparing against baselines using bag-of- Bridge over the abyss words and bag-of-bigrams for state rep- resentations. Our algorithm outperforms Figure 1: Sample gameplay from a Fantasy World. the baselines on both worlds demonstrat- The player with the quest of finding a secret tomb, ing the importance of learning expressive is currently located on an old bridge. She then 1 representations. chooses an action to go east that brings her to a ruined gatehouse (State 2). 1 Introduction challenging for existing AI programs to play these In this paper, we address the task of learning con- games (DePristo and Zubek, 2001). trol policies for text-based strategy games. These In designing an autonomous game player, we games, predecessors to modern graphical ones, have considerable latitude when selecting an ad- still enjoy a large following worldwide.2 They of- equate state representation to use. The simplest ten involve complex worlds with rich interactions method is to use a bag-of-words representation and elaborate textual descriptions of the underly- derived from the text description. However, this ing states (see Figure 1). Players read descriptions scheme disregards the ordering of words and the of the current world state and respond with natural finer nuances of meaning that evolve from com- language commands to take actions. Since the un- posing words into sentences and paragraphs. For derlying state is not directly observable, the player instance, in State 2 in Figure 1, the agent has to has to understand the text in order to act, making it understand that going east will lead it to the cas- tle whereas moving south will take it to the stand- ∗ Both authors contributed equally to this work. ing archway. An alternative approach is to convert 1Code is available at http://people.csail.mit. edu/karthikn/mud-play. text descriptions to pre-specified representations 2http://mudstats.com/ using annotated training data, commonly used in language grounding tasks (Matuszek et al., 2013; and playing computer games (Eisenstein et al., Kushman et al., 2014). 2009; Branavan et al., 2011a). In contrast, our goal is to learn useful represen- Games provide a rich domain for grounded lan- tations in conjunction with control policies. We guage analysis. Prior work has assumed perfect adopt a reinforcement learning framework and for- knowledge of the underlying state of the game to mulate game sequences as Markov Decision Pro- learn policies. Gorniak and Roy (2005) developed cesses. An agent playing the game aims to maxi- a game character that can be controlled by spoken mize rewards that it obtains from the game engine instructions adaptable to the game situation. The upon the occurrence of certain events. The agent grounding of commands to actions is learned from learns a policy in the form of an action-value func- a transcript manually annotated with actions and tion Q(s; a) which denotes the long-term merit of state attributes. Eisenstein et al. (2009) learn game an action a in state s. rules by analyzing a collection of game-related The action-value function is parametrized us- documents and precompiled traces of the game. In ing a deep recurrent neural network, trained us- contrast to the above work, our model combines ing the game feedback. The network contains two text interpretation and strategy learning in a single modules. The first one converts textual descrip- framework. As a result, textual analysis is guided tions into vector representations that act as prox- by the received control feedback, and the learned ies for states. This component is implemented us- strategy directly builds on the text interpretation. ing Long Short-Term Memory (LSTM) networks Our work closely relates to an automatic game (Hochreiter and Schmidhuber, 1997). The second player that utilizes text manuals to learn strategies module of the network scores the actions given the for Civilization (Branavan et al., 2011a). Similar vector representation computed by the first. to our approach, text analysis and control strate- We evaluate our model using two Multi-User gies are learned jointly using feedback provided Dungeon (MUD) games (Curtis, 1992; Amir and by the game simulation. In their setup, states are Doyle, 2002). The first game is designed to pro- fully observable, and the model learns a strategy vide a controlled setup for the task, while the sec- by combining state/action features and features ond is a publicly available one and contains hu- extracted from text. However, in our application, man generated text descriptions with significant the state representation is not provided, but has to language variability. We compare our algorithm be inferred from a textual description. Therefore, against baselines of a random player and mod- it is not sufficient to extract features from text to els that use bag-of-words or bag-of-bigrams rep- supplement a simulation-based player. resentations for a state. We demonstrate that our Another related line of work consists of auto- model LSTM-DQN significantly outperforms the matic video game players that infer state repre- baselines in terms of number of completed quests sentations directly from raw pixels (Koutn´ık et al., and accumulated rewards. For instance, on a fan- 2013; Mnih et al., 2015). For instance, Mnih et tasy MUD game, our model learns to complete al. (2015) learn control strategies using convolu- 96% of the quests, while the bag-of-words model tional neural networks, trained with a variant of and a random baseline solve only 82% and 5% of Q-learning (Watkins and Dayan, 1992). While the quests, respectively. Moreover, we show that both approaches use deep reinforcement learning the acquired representation can be reused across for training, our work has important differences. games, speeding up learning and leading to faster In order to handle the sequential nature of text, we convergence of Q-values. use Long Short-Term Memory networks to auto- matically learn useful representations for arbitrary 2 Related Work text descriptions. Additionally, we show that de- composing the network into a representation layer Learning control policies from text is gaining in- and an action selector is useful for transferring the creasing interest in the NLP community. Example learnt representations to new game scenarios. applications include interpreting help documenta- tion for software (Branavan et al., 2010), navi- 3 Background gating with directions (Vogel and Jurafsky, 2010; Kollar et al., 2010; Artzi and Zettlemoyer, 2013; Game Representation We represent a game by Matuszek et al., 2013; Andreas and Klein, 2015) the tuple hH; A; T; R; Ψi, where H is the set of all possible game states, A = f(a; o)g is the set of Q(s, a) Q(s, o) all commands (action-object pairs), T (h0 j h; a; o) is the stochastic transition function between states Linear Linear ( ) and R h; a; o is the reward function. The game ReLU φA state H is hidden from the player, who only re- ceives a varying textual description, produced by Linear a stochastic function Ψ: H ! S. Specifically, vs the underlying state h in the game engine keeps Mean Pooling track of attributes such as the player’s location, her health points, time of day, etc. The function LSTM LSTM LSTM LSTM φR Ψ (also part of the game framework) then converts w1 w2 w3 wn this state into a textual description of the location the player is at or a message indicating low health. We do not assume access to either H or Ψ for our Figure 2: Architecture of LSTM-DQN: The Rep- agent during both training and testing phases of resentation Generator (φR) (bottom) takes as input our experiments. We denote the space of all possi- a stream of words observed in state s and produces ble text descriptions s to be S. Rewards are gener- a vector representation vs, which is fed into the ated using R and are only given to the player upon action scorer (φA) (top) to produce scores for all completion of in-game quests. actions and argument objects. Q-Learning Reinforcement Learning is a com- state-action pairs. One solution to this problem monly used framework for learning control poli- is to approximate Q(s; a) using a parametrized cies in game environments (Silver et al., 2007; function Q(s; a; θ), which can generalize over Amato and Shani, 2010; Branavan et al., 2011b; states and actions by considering higher-level at- Szita, 2012).
Recommended publications
  • Planning for Non-Player Characters by Learning from Demonstration
    University of Pennsylvania ScholarlyCommons Publicly Accessible Penn Dissertations 2018 Planning For Non-Player Characters By Learning From Demonstration John Drake University of Pennsylvania, [email protected] Follow this and additional works at: https://repository.upenn.edu/edissertations Part of the Computer Sciences Commons Recommended Citation Drake, John, "Planning For Non-Player Characters By Learning From Demonstration" (2018). Publicly Accessible Penn Dissertations. 2756. https://repository.upenn.edu/edissertations/2756 This paper is posted at ScholarlyCommons. https://repository.upenn.edu/edissertations/2756 For more information, please contact [email protected]. Planning For Non-Player Characters By Learning From Demonstration Abstract In video games, state of the art non-player character (NPC) behavior generation typically depends on hard-coding NPC actions. In many game situations however, it is hard to foresee how an NPC should behave to appear intelligent or to accommodate human preferences for NPC behavior. We advocate the creation of a more flexible method ot allow players (and developers) to train NPCs to execute novel behaviors which are not hard-coded. In particular, we investigate search-based planning approaches using demonstration to guide the search through high-dimensional spaces that represent the full state of the game. To this end, we developed the Training Graph heuristic, an extension of the Experience Graph heuristic, that guides a search smoothly and effectively even when a demonstration is unreachable in the search space, and ensures that more of the demonstrations are utilized to better train the NPC's behavior. To deal with variance in the initial conditions of such planning problems, we have developed heuristics in the Multi-Heuristic A* framework to adapt demonstration trace data to new problems.
    [Show full text]
  • Improving the Social Believability of Non-Player Characters in Role-Playing Games
    Agents that Relate: Improving the Social Believability of Non-Player Characters in Role-Playing Games Nuno Afonso and Rui Prada IST-Technical University of Lisbon, INESC-ID Avenida Prof. Cavaco Silva – TagusPark 2744-016 Porto Salvo, Portugal {nafonso, rui.prada}@gaips.inesc-id.pt Abstract. As the video games industry grows and video games become more part of our lives, we are eager for better gaming experiences. One field in which games still have much to gain is in Non-Player Character behavior in socially demanding games, like Role-Playing Games. In Role-Playing Games players have to interact constantly with very simple Non-Player Characters, with no social behavior in most of the cases, which contrasts with the rich social experience that was provided in its traditional pen-and-paper format. What we propose in this paper is that if we create a richer social behavior in Non-Player Characters the player’s gaming experience can be improved. In order to attain this we propose a model that has at its core social relationships with/between Non-Player Characters. By doing an evaluation with players, we identified that 80% of them preferred such system, affirming that it created a better gaming experience. Keywords: role-playing games, non-player characters, artificial intelligence, social, behavior, relationship, personality, theory of mind. 1 Introduction As computers became more present in our lives, they transformed from simple work machines into a means of entertainment. This created a massive industry of video games, an industry that beats sales records every year, and has a yearly growth rate higher than any other entertainment industry [2].
    [Show full text]
  • Making Quests Playable with Computer–Assisted
    UC Irvine Cognition and Creativity Title QuestBrowser: Making Quests Playable with Computer- Assisted Design Permalink https://escholarship.org/uc/item/2tk0h882 Authors Sullivan, Anne Mateas, Michael Wardrip–Fruin, Noah Publication Date 2009-12-12 Peer reviewed eScholarship.org Powered by the California Digital Library University of California QuestBrowser: Making Quests Playable with Computer- Assisted Design Anne Sullivan Michael Mateas Noah Wardrip-Fruin Expressive Intelligence Studio Expressive Intelligence Studio Expressive Intelligence Studio University of California, Santa Cruz University of California, Santa Cruz University of California, Santa Cruz [email protected] [email protected] [email protected] ABSTRACT narrative; within which the player moves through the experience, Good gameplay has been characterized as a series of interesting fulfilling check points to advance the story. These story choices. Therefore, to have gameplay of any sort requires the checkpoints are delivered to the player in the form of quests, and player to be presented with decisions. Given this definition, many these quests lack interesting or meaningful choices, but are instead quests within computer role-playing games are not playable as delivered as required actions for the player to complete. they currently exist. Instead, quests are given to the player as a Because of this lack of choice, we argue that quests are not series of tasks to perform in a specific way in order to advance the currently playable in many CRPGs. When we look at table-top story within the game. We look at making quests playable – role-playing games, the precursor to CRPGs, we find quests that adding choices for the player – and what a system that could exist in their playable form and are used with regularity.
    [Show full text]
  • Dynamic Quest Generation Using in Session Player Behaviors in MMORPG
    WRITING FOR EACH OTHER: DYNAMIC QUEST GENERATION USING IN SESSION PLAYER BEHAVIORS IN MMORPG A Thesis presented to the Faculty of California Polytechnic State University, San Luis Obispo In Partial Fulfillment of the Requirements for the Degree Master of Science in Computer Science by Sean Mendonca June 2020 c 2020 Sean Mendonca ALL RIGHTS RESERVED ii COMMITTEE MEMBERSHIP TITLE: Writing For Each Other: Dynamic Quest Generation Using In Session Player Behav- iors in MMORPG AUTHOR: Sean Mendonca DATE SUBMITTED: June 2020 COMMITTEE CHAIR: Foaad Khosmood, Ph.D. Professor of Computer Science COMMITTEE MEMBER: Michael Haungs, Ph.D. Professor of Computer Science COMMITTEE MEMBER: Franz Kurfess, Ph.D. Professor of Computer Science iii ABSTRACT Writing For Each Other: Dynamic Quest Generation Using In Session Player Behaviors in MMORPG Sean Mendonca Role-playing games (RPGs) rely on interesting and varied experiences to maintain player attention. These experiences are often provided through quests, which give players tasks that are used to advance stories or events unfolding in the game. Tra- ditional quests in video games require very specific conditions to be met, and for participating members to advance them by carrying out pre-defined actions. These types of quests are generated with perfect knowledge of the game world and are able to force desired behaviors out of the relevant non-player characters (NPCs). This becomes a major issue in massive multiplayer online (MMO) when other players can often disrupt the conditions needed for quests to unfold in a believable and immersive way, leading to the absence of a genuine multiplayer RPG experience. Our proposed solution is to dynamically create quests from real-time information on the unscripted actions of other NPCs and players in a game.
    [Show full text]
  • Osrs Quest List Order
    Osrs Quest List Order Coyish Brett sometimes refining his overhaul shrinkingly and capitalize so ornamentally! Diaphanous Micky anneal or knownunscrambling bumptiously some orpotoroos items equanimously sexennially, however when interneural trifoliate ReedSanford sunk sledge-hammers entomologically dialectically or conjecturing. and genealogically.Christofer usually Use any particular list! This number the most usually order to minimize leveling done outside the quest reward xp Other mortal being a. Subscribe for a bit with related posts you may be alchemised psa: a vial of its a bunch of. Making guide describes unlocking an account, simply the best in our official legal and go, including ip addresses, potentially give it! Welcome to anvils after that guides you order, and social media feed, as a free in quest osrs list order for quick teleport. Players can be listed in? Ask your order from another quest order! Guarantee you order has always restored a list of fish found in the osrs bank automatically embedded. Click more information shall be nice day after struggling for free. Use any order to be listed in an account is not include hyperlink to. Each skill capes for gold bar that list from the magic bonuses are the ability from divine bastion can also convert traditional slime pit and purchase. It contains the primary ingredient of tough quests? Mozilla using cooking recipes are anvils here is a whole lot of guthix rest of page you need a dedicated team is a title that. Where players in. That list of osrs this maze are listed in order: dollars for staff makes it allows you did fish breed profile.
    [Show full text]
  • Free to Play Recommended First Quests Runescape
    Free To Play Recommended First Quests Runescape Opisthognathous Regan clog that rheology cross under and illustrating acrogenously. Reggy is almighty shroud-laid after powerful detested.Sherwood preside his peroneus agilely. Spinning Bartel lowns no nails remerged irksomely after Ashish engrains unsympathetically, quite Website uses in old school bond in that this game will pay in old school bond hack has runescape play it to get the rum deal with Wizard robes as you by jogres and can. General Store, triple the coming years. These apps are designed to improve each game experiance by pushing the bounderies of your browser. If we want. Shulk: You lost the funniest things. They reach not reset for fashion week. Same story wise since the observatory reception room back to accommodate thicker socks and talk to get started, these must exactly? These free play. We bail you retire use our commenting platform to engage in insightful conversations about issues in whose community. Why is recommended quests, quest for late into the same class, with a wall of nintendo switch. She will secure you to reed the house cannot save him seize the other servants. Tell you play runescape quest cape for easier to. They wax the hard step of producing potions, except for potions brewed with coconut milk. We can play runescape quests are free linux games are perfect for easier once bigg boss is recommended, questing experience as. Secure stitch Your Devices. We were looking for racers or if any other related to help shazam prot warrior, log in mourning after in the range of gp for! This quest cape of quests and play on your first things to teleport you to fire on a very own unique welcome to my phone.
    [Show full text]
  • Mud Connector
    Archive-name: mudlist.doc /_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ /_/_/_/_/ THE /_/_/_/_/ /_/_/ MUD CONNECTOR /_/_/ /_/_/_/_/ MUD LIST /_/_/_/_/ /_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ o=======================================================================o The Mud Connector is (c) copyright (1994 - 96) by Andrew Cowan, an associate of GlobalMedia Design Inc. This mudlist may be reprinted as long as 1) it appears in its entirety, you may not strip out bits and pieces 2) the entire header appears with the list intact. Many thanks go out to the mud administrators who helped to make this list possible, without them there is little chance this list would exist! o=======================================================================o This list is presented strictly in alphabetical order. Each mud listing contains: The mud name, The code base used, the telnet address of the mud (unless circumstances prevent this), the homepage url (if a homepage exists) and a description submitted by a member of the mud's administration or a person approved to make the submission. All listings derived from the Mud Connector WWW site http://www.mudconnect.com/ You can contact the Mud Connector staff at [email protected]. [NOTE: This list was computer-generated, Please report bugs/typos] o=======================================================================o Last Updated: June 8th, 1997 TOTAL MUDS LISTED: 808 o=======================================================================o o=======================================================================o Muds Beginning With: A o=======================================================================o Mud : Aacena: The Fatal Promise Code Base : Envy 2.0 Telnet : mud.usacomputers.com 6969 [204.215.32.27] WWW : None Description : Aacena: The Fatal Promise: Come here if you like: Clan Wars, PKilling, Role Playing, Friendly but Fair Imms, in depth quests, Colour, Multiclassing*, Original Areas*, Tweaked up code, and MORE! *On the way in The Fatal Promise is a small mud but is growing in size and player base.
    [Show full text]
  • Quest Guidebook 2021
    Guidebook Overview We are so thrilled to welcome your child to a Tawonga Quest this summer! This 2021 Quest Guidebook has everything you need to know to get ready for your adventure, including helpful information on: ❖ COVID-19 Policies ❖ Screen-Free Policy ❖ No-Package Policy ❖ Drop Off and Return Information ❖ Mail Information ❖ What to Bring and What Not to Bring ❖ Food Including Allergies and Dietary Restrictions ❖ Behavior Expectations ❖ And Much More! Please help prepare your child for a safe and happy summer by reviewing this guidebook and discussing it with your child before their trip. Keep This Guidebook The information it contains is updated annually and will be useful before and throughout the summer, whether this is your family’s first summer participating in a Tawonga program or you have attended for many years. The guidebook is also available on our website at tawonga.org under the “Forms and Resources” section. If you have any questions or concerns, please call us at 415.543.2267 or email [email protected]. After you finish reading this guidebook, please sign the authorization and release forms for your Quest. 1 The Family | Tawonga Partnership Congratulations! You’ve registered for a great summer program with Camp Tawonga. We hope to make this summer a positive experience, and your help is essential. Reading this manual is a key step in preparing for your adventure ahead. This guidebook is intended to be read both by teens participating in a Quest and by their parents. Whether your family is new or returning, please carefully read this guidebook, discuss its contents together, and save it for future reference.
    [Show full text]
  • Causes, Magnitude and Implications of Griefing in Massively Multiplayer Online Role-Playing Games
    Causes, Magnitude and Implications of Griefing in Massively Multiplayer Online Role-Playing Games LEIGH SIMON ACHTERBOSCH This thesis is submitted in total fulfilment of the requirements for the degree of Doctor of Philosophy (Information Technology) Faculty of Science and Technology Federation University Australia PO Box 663 University Drive, Mount Helen Ballarat, Victoria 3353 Australia Submitted in April 2015 Abstract Abstract This thesis presents findings from research into the global phenomenon known as griefing that occurs in Massively Multiplayer Online Role-Playing Games (MMORPGs). Griefing, in its simplest terms, refers to the act of one player intentionally disrupting another player’s game experience for personal pleasure and potential gain. For too long it has been unknown how pervasive griefing is, how frequently griefing occurs and, in particular, the impact on players that are subjected to griefing. There has also been limited research regarding what causes a player to perform griefing. This thesis addresses these concerns by answering the research question “What are the causes and implications of griefing in Massively Multiplayer Online Role-Playing Games, and what magnitude of griefing exists in this genre?” Mixed method research was employed using the “Sequential Explanatory Strategy”, in which a quantitative phase was followed by a qualitative phase to strengthen the findings. The quantitative phase consisted of a survey that attracted 1188 participants of a representative player population. The qualitative phase consisted of interviews with 15 participants to give more personalised data. The data was analysed from the perspectives of different demographics and different associations to griefing. The thesis contributed original findings regarding the causes, magnitude and implications of griefing in MMORPGs.
    [Show full text]
  • NEXON M and Socialspiel Entertainment Reveal Legacy Quest
    Feb. 26, 2015 NEXON Co., Ltd. NEXON M and Socialspiel Entertainment Reveal Legacy Quest A Roguelike Dungeon Crawling Action RPG Coming Soon for Mobile Devices TOKYO – February 26, 2015 – NEXON Co., Ltd. (“Nexon”) (3659.TO), a worldwide leader in free-to-play online games, today announced that its mobile game publishing subsidiary, NEXON M Inc. (“NEXON M”) and a Vienna, Austria-based mobile games studio, Socialspiel Entertainment GmbH (“Socialspiel”), will release Legacy Quest, an upbeat, roguelike-inspired action role playing game (RPG) for mobile devices. Legacy Quest puts players in the fearless boots of a proud line of customizable heroes, where they will explore massive dungeons while fighting monsters and uncovering rare loot. Death of a hero is permanent in Legacy Quest, but every creature vanquished and Soul collected will give players the ability to unlock powerful new traits, enhancing the next hero in their bloodline to continue their epic adventure and take on increasingly challenging monsters and dungeons. Players will adventure through ever-evolving dungeons set in a unique and highly stylized fantasy universe, inspired by the classic 8-bit RPGs of previous generations. 1 “Socialspiel is taking a fresh approach to the fantasy RPG genre with Legacy Quest’s unique customization and crafting, and a vibrant, gorgeous art style,” said John Robinson, General Manager of NEXON M, Nexon’s mobile game publishing division. Players will create, customize and upgrade their weapons to fit their play style through Legacy Quest’s unique crafting and customization system. If a hero is slain in a dungeon, their custom gear will be added as new items to the game world and become discoverable by other players as they hack and slash their way through waves of monsters.
    [Show full text]
  • Making Quests Playable with Computer–Assisted
    QuestBrowser: Making Quests Playable with Computer- Assisted Design Anne Sullivan Michael Mateas Noah Wardrip-Fruin Expressive Intelligence Studio Expressive Intelligence Studio Expressive Intelligence Studio University of California, Santa Cruz University of California, Santa Cruz University of California, Santa Cruz [email protected] [email protected] [email protected] ABSTRACT narrative; within which the player moves through the experience, Good gameplay has been characterized as a series of interesting fulfilling check points to advance the story. These story choices. Therefore, to have gameplay of any sort requires the checkpoints are delivered to the player in the form of quests, and player to be presented with decisions. Given this definition, many these quests lack interesting or meaningful choices, but are instead quests within computer role-playing games are not playable as delivered as required actions for the player to complete. they currently exist. Instead, quests are given to the player as a Because of this lack of choice, we argue that quests are not series of tasks to perform in a specific way in order to advance the currently playable in many CRPGs. When we look at table-top story within the game. We look at making quests playable – role-playing games, the precursor to CRPGs, we find quests that adding choices for the player – and what a system that could exist in their playable form and are used with regularity. support playable quests would look like. Finally, we address the impact playable quests would have on a designer and discuss In this paper, we will describe our solution for this shortcoming of QuestBrowser, the system we created to handle these concerns.
    [Show full text]
  • Improving the Social Believability of Non-Player Characters in Role-Playing Games
    Agents That Relate: Improving the Social Believability of Non-Player Characters in Role-Playing Games Nuno Afonso and Rui Prada IST-Technical University of Lisbon, INESC-ID Avenida Prof. Cavaco Silva – TagusPark 2744-016 Porto Salvo, Portugal {nafonso,rui.prada}@gaips.inesc-id.pt Abstract. As the video games industry grows and video games become more part of our lives, we are eager for better gaming experiences. One field in which games still have much to gain is in Non-Player Character behavior in socially demanding games, like Role-Playing Games. In Role-Playing Games players have to interact constantly with very simple Non-Player Characters, with no social behavior in most of the cases, which contrasts with the rich social experi- ence that was provided in its traditional pen-and-paper format. What we pro- pose in this paper is that if we create a richer social behavior in Non-Player Characters the player’s gaming experience can be improved. In order to attain this we propose a model that has at its core social relationships with/between Non-Player Characters. By doing an evaluation with players, we identified that 80% of them preferred such system, affirming that it created a better gaming experience. Keywords: role-playing games, non-player characters, artificial intelligence, social, behavior, relationship, personality, theory of mind. 1 Introduction As computers became more present in our lives, they transformed from simple work machines into a means of entertainment. This created a massive industry of video games, an industry that beats sales records every year, and has a yearly growth rate higher than any other entertainment industry [2].
    [Show full text]