ACT grew out of and semantic nets. It differs from other cognitive research on human memory. Over the years, architectures in that it also includes signals, ACT has evolved into ACT* and more recently, images, and maps in its knowledge database, ACT-R. ACT-R is being used in several research and maintains a tight real-time coupling projects in an Advanced Decision Architectures between iconic and symbolic data structures in Collaborative Technology Alliance for the U.S. its world model. The 4D/RCS architecture is also Army (Gonzalez 2003). ACT-R is also being different in its (1) focus on task decomposition used by thousands of schools across the country as the fundamental organizing principle; as an algebra tutor--an instructional system (2) level of specificity in the assignment of duties that supports learning by doing. Another wellknown and responsibilities to agents and units in and widely used architecture is Soar the behavior-generating hierarchy; and (3) emphasis (Laird, Newell, and Rosenbloom 1987). Soar on controlling real machines in realworld grew out of research on human problem solving environments.

We are developing Companion Cognitive Systems, a new kind of software that can be effectively treated as a collaborator. Aside from their potential utility, we believe this effort is important because it focuses on three key problems that must be solved to achieve human-level AI: Robust reasoning and learning, interactivity, and longevity. We describe the ideas we are using to develop the first architecture for Companions: analogical processing, grounded in cognitive science for reasoning and learning, sketching and concept maps to improve interactivity, and a distributed agent architecture hosted on a cluster to achieve performance and longevity. We outline some results on learning by accumulating examples derived from our first experimental version.

Making progress toward human-level artificial intelligence often seems to require a large number of difficult-to-integrate computational methods and enormous amounts of knowledge about the world. This article provides evidence from linguistics, cognitive psychology, and neuroscience for the cognitive substrate hypothesis that a relatively small set of properly integrated data structures and algorithms can underlie the whole range of cognition required for human-level intelligence. Some computational principles (embodied in the Polyscheme cognitive architecture) are proposed to solve the integration problems involved in implementing such a substrate. A natural language syntactic parser that uses only the mechanisms of an infant physical reasoning model developed in Polyscheme demonstrates that a single cognitive substrate can underlie intelligent systems in superficially very dissimilar domains. This work suggests that identifying and implementing a cognitive substrate will accelerate progress toward human-level artificial intelligence.

In this article, I claim that research on cognitive architectures is an important path to the development of general intelligent systems. I contrast this paradigm with other approaches to constructing such systems, and I review the theoretical commitments associated with a cognitive architecture. I illustrate these ideas using a particular architecture -- ICARUS -- by examining its claims about memories, about the representation and organization of knowledge, and about the performance and learning mechanisms that affect memory structures. I also consider the high-level programming language that embodies these commitments, drawing examples from the domain of in-city driving. In closing, I consider ICARUS's relation to other cognitive architectures and discuss some open issues that deserve increased attention.

In recent years, CP-nets have emerged as a useful tool for supporting preference elicitation, reasoning, and representation. CP-nets capture and support reasoning with qualitative conditional preference statements, statements that are relatively natural for users to express. In this paper, we extend the CP-nets formalism to handle another class of very natural qualitative statements one often uses in expressing preferences in daily life - statements of relative importance of attributes. The resulting formalism, TCP-nets, maintains the spirit of CP-nets, in that it remains focused on using only simple and natural preference statements, uses the ceteris paribus semantics, and utilizes a graphical representation of this information to reason about its consistency and to perform, possibly constrained, optimization using it. The extra expressiveness it provides allows us to better model tradeoffs users would like to make, more faithfully representing their preferences.

In a peer-to-peer inference system, each peer can reason locally but can also solicit some of its acquaintances, which are peers sharing part of its vocabulary. In this paper, we consider peer-to-peer inference systems in which the local theory of each peer is a set of propositional clauses defined upon a local vocabulary. An important characteristic of peer-to-peer inference systems is that the global theory (the union of all peer theories) is not known (as opposed to partition-based reasoning systems). The main contribution of this paper is to provide the first consequence finding algorithm in a peer-to-peer setting: DeCA. It is anytime and computes consequences gradually from the solicited peer to peers that are more and more distant. We exhibit a sufficient condition on the acquaintance graph of the peer-to-peer inference system for guaranteeing the completeness of this algorithm. Another important contribution is to apply this general distributed reasoning setting to the setting of the Semantic Web through the Somewhere semantic peer-to-peer data management system. The last contribution of this paper is to provide an experimental analysis of the scalability of the peer-to-peer infrastructure that we propose, on large networks of 1000 peers.

The treatment of exogenous events in planning is practically important in many real-world domains where the preconditions of certain plan actions are affected by such events. In this paper we focus on planning in temporal domains with exogenous events that happen at known times, imposing the constraint that certain actions in the plan must be executed during some predefined time windows. When actions have durations, handling such temporal constraints adds an extra difficulty to planning. We propose an approach to planning in these domains which integrates constraint-based temporal reasoning into a graph-based planning framework using local search. Our techniques are implemented in a planner that took part in the 4th International Planning Competition (IPC-4). A statistical analysis of the results of IPC-4 demonstrates the effectiveness of our approach in terms of both CPU-time and plan quality. Additional experiments show the good performance of the temporal reasoning techniques integrated into our planner.

We study an approach to policy selection for large relational Markov Decision Processes (MDPs). We consider a variant of approximate policy iteration (API) that replaces the usual value-function learning step with a learning step in policy space. This is advantageous in domains where good policies are easier to represent and learn than the corresponding value functions, which is often the case for the relational MDPs we are interested in. In order to apply API to such problems, we introduce a relational policy language and corresponding learner. In addition, we introduce a new bootstrapping routine for goal-based planning domains, based on random walks. Such bootstrapping is necessary for many large relational MDPs, where reward is extremely sparse, as API is ineffective in such domains when initialized with an uninformed policy. Our experiments show that the resulting system is able to find good policies for a number of classical planning domains and their stochastic variants by solving them as extremely large relational MDPs. The experiments also point to some limitations of our approach, suggesting future work.

A decision process in which rewards depend on history rather than merely on the current state is called a decision process with non-Markovian rewards (NMRDP). In decision-theoretic planning, where many desirable behaviours are more naturally expressed as properties of execution sequences rather than as properties of states, NMRDPs form a more natural model than the commonly adopted fully Markovian decision process (MDP) model. While the more tractable solution methods developed for MDPs do not directly apply in the presence of non-Markovian rewards, a number of solution methods for NMRDPs have been proposed in the literature. These all exploit a compact specification of the non-Markovian reward function in temporal logic, to automatically translate the NMRDP into an equivalent MDP which is solved using efficient MDP solution methods. This paper presents NMRDPP (Non-Markovian Reward Decision Process Planner), a software platform for the development and experimentation of methods for decision-theoretic planning with non-Markovian rewards. The current version of NMRDPP implements, under a single interface, a family of methods based on existing as well as new approaches which we describe in detail. These include dynamic programming, heuristic search, and structured methods. Using NMRDPP, we compare the methods and identify certain problem features that affect their performance. NMRDPP's treatment of non-Markovian rewards is inspired by the treatment of domain-specific search control knowledge in the TLPlan planner, which it incorporates as a special case. In the First International Probabilistic Planning Competition, NMRDPP was able to compete and perform well in both the domain-independent and hand-coded tracks, using search control knowledge in the latter.

The emerging Semantic Web focuses on bringing knowledge representationlike capabilities to Web applications in a Web-friendly way. The ability to put knowledge on the Web, share it, and reuse it through standard Web mechanisms provides new and interesting challenges to artificial intelligence. In this paper, I explore the similarities and differences between the Semantic Web and traditional AI knowledge representation systems, and see if I can validate the analogy "The Semantic Web is to KR as the Web is to hypertext."