Previous research demonstrates that social experience is a key motivation for continued play in online games. We argue that there is an untapped potential for NPCs to act as social mediators in online games by stimulating social interaction between players. For example, a social mediator NPC could ask in chat, “How is everyone’s day?” to encourage conversations between players. We suggest two potential applications for social mediator NPCs in online games by examining social experiences in two contemporary massively-multiplayer online role-playing games (MMOs): Guild Wars 2 and World of Warcraft.

This paper presents a model of exaggeration suitable for implementation on a robot. Exaggeration is an interest form of dishonesty in that it serves as a tradeoff between the different costs associated with lying and the reward received by having one’s lie accepted. Moreover, exaggeration offers the deceiver additional control in the form of much the exaggerated statement differs from the truth. We use a color guessing game to examine the different tradeoffs between these costs and rewards and their impact on exaggeration. Our results indicate some amount of exaggeration is the preferred option during most early interactions. Further, because the cost of lying increases linear with the number of lies, exaggeration decreases with additional interactions. We conclude by arguing why social robots must be capable of lying.

An accurate problem list plays the key role of a problem-oriented medical record, which plays a significant role in improving patient care. However, the multi-author, multi-purpose nature of problem list makes it a challenge to maintain, and a single list is difficult, if not impossible, to satisfy all the needs of different practitioners. In this paper, we propose using machine generated problem list to assist a medical practitioner to review a patient’s chart. The proposed system scans both structured and unstructured data in a patient’s electronic medical record (EMR) and generates a ranked, recall-oriented problem list grouped by body systems. Details of each problem are readily available for the user to assess the correctness and relevance of the problem. The user can then provide feedback to the system on the trustworthiness of each evidence passage retrieved, as well as the validity of the problem as a whole. The user-specific feedback provides new information the system needs to perform active learning to learn the user’s preference and produce personalized, and/or domain-specific problem lists.

Game tree search algorithms, such as Monte Carlo Tree Search (MCTS), require access to a forward model (or "simulator") of the game at hand. However, in some games such forward model is not readily available. In this paper we address the problem of automatically learning forward models (more specifically, combats models) for two-player attrition games. We report experiments comparing several approaches to learn such combat model from replay data to models generated by hand. We use StarCraft, a Real-Time Strategy (RTS) game, as our application domain. Specifically, we use a large collection of already collected replays, and focus on learning a combat model for tactical combats.

Is it possible to conduct player modeling without any players? In this paper we use Monte-Carlo Tree Search-controlled procedural personas to simulate a range of decision making styles in the puzzle game MiniDungeons 2. The purpose is to provide a method for synthetic play testing of game levels with synthetic players based on designer intuition and experience. Five personas are constructed, representing five different decision making styles archetypal for the game. The personas vary solely in the weights of decision-making utilities that describe their valuation of a set affordances in MiniDungeons 2. By configuring these weights using designer expert knowledge, and passing the configurations directly to the MCTS algorithm, we make the personas exhibit a number of distinct decision making and play styles.

Coordination with an unknown human teammate is a notable challenge for cooperative agents. Behavior of human players in games with cooperating AI agents is often sub-optimal and inconsistent leading to choreographed and limited cooperative scenarios in games. This paper considers the difficulty of cooperating with a teammate whose goal and corresponding behavior change periodically. Previous work uses Bayesian models for updating beliefs about cooperating agents based on observations. We describe belief models for on-line planning, discuss tuning in the presence of noisy observations, and demonstrate empirically its effectiveness in coordinating with inconsistent agents in a simple domain. Further work in this area promises to lead to techniques for more interesting cooperative AI in games.

Interactive narrative systems produce branching story experiences for a human user using an interactive world. A class of interactive narrative systems, called strong story systems, manage a user's experience by manipulating the interactive world and its characters according to a formal story model. In these systems, a human user may place the world into a state such that the formal story model can no longer control interaction. One solution to this problem, called intervention, is to exchange the undesirable outcomes of a player's action for a set that do not violate the story model. However, the player may become aware that their intended action is being intervened against by a context-sensitive, meta-narrative process. In this paper we describe a method of ensuring game world alibis for interventions through domain modification of world mechanics.

Game designers can use computer-aided game design methods to quantitatively compare player skill levels, different game variants, and learning rates, for the purpose of modeling how players will likely experience a game. We use Monte-Carlo simulation, hazard functions, and survival analysis to show how difficulty will quantitatively change throughout a game level as we vary skill, game parameters, and learning rates. We give a mathematical overview of survival analysis, present empirical data analyses of our player models for each game variant, and provide theoretical probability distributions for each game. This analysis shows the quantitative reasons why balancing a game for a wide range of player skill can be difficult; our player modeling provides tools for tuning this game balance. We also analyze the score distribution of over 175 million play sessions of a popular online Flappy Bird variant to demonstrate how learning effects can impact scores, implying that learning is crucial aspect of player modeling.

Natural language dialogue provides the opportunity for truly natural human-robot interaction. A robot participating in natural language dialogue must identify or create new representations for referenced entities if it is to discuss, reason about, or perform actions involving that entity, a capability known as reference resolution. In previous work we presented algorithms for resolving references occurring in definite noun phrases. In this paper we propose an algorithm for resolving references in a wider array of linguistic forms, using the Givenness Hierarchy.

The Markov Decision Process (MDP) framework is a versatile method for addressing single and multiagent sequential decision making problems. Many exact and approximate solution methods attempt to exploit structure in the problem and are based on value factorization. Especially multiagent settings (MAS), however, are known to suffer from an exponential increase in value component sizes as interactions become denser, meaning that approximation architectures are overly restricted in the problem sizes and types they can handle. We present an approach to mitigate this limitation for certain types of MASs, exploiting a property that can be thought of as "anonymous influence" in the factored MDP. In particular, we show how anonymity can lead to representational and computational efficiencies, both for general variable elimination in a factor graph but also for the approximate linear programming solution to factored MDPs. The latter allows to scale linear programming to factored MDPs that were previously unsolvable. Our results are shown for a disease control domain over a graph with 50 nodes that are each connected with up to 15 neighbors.