In the framework of fully cooperative multi-agent systems, independent (non-communicative) agents that learn by reinforcement must overcome several difficulties to manage to coordinate. This paper identifies several challenges responsible for the non-coordination of independent agents: Pareto-selection, non-stationarity, stochasticity, alter-exploration and shadowed equilibria. A selection of multi-agent domains is classified according to those challenges: matrix games, Boutilier's coordination game, predators pursuit domains and a special multi-state game. Moreover, the performance of a range of algorithms for independent reinforcement learners is evaluated empirically. Those algorithms are Q-learning variants: decentralized Q-learning, distributed Q-learning, hysteretic Q-learning, recursive frequency maximum Q-value and win-or-learn fast policy hill climbing. An overview of the learning algorithms’ strengths and weaknesses against each challenge concludes the paper and can serve as a basis for choosing the appropriate algorithm for a new domain. Furthermore, the distilled challenges may assist in the design of new learning algorithms that overcome these problems and achieve higher performance in multi-agent applications.

In recent years, digital content of three-dimensional (3D) scientific objects has become widespread and is made available in a plethora of on-line scientific repositories. A systematic and formal approach becomes necessary to represent the knowledge/information related to these objects, in order to facilitate their search, retrieval and reuse. In this paper, we employ semantic interoperability to manage 3D scientific object content in a large-scale framework of scientific applications that consists of 3D shape models, associated tools and resources. We use an advanced ontological organization for their metadata, powerful reasoning engines for their search and retrieval and we elucidate several crucial issues in the design and implementation of 3D knowledge-based management systems.

This paper describes and compares the most relevant approaches for associating tags with semantics in order to make explicit the meaning of those tags. We identify a common set of steps that are usually considered across all these approaches and frame our descriptions according to them, providing a unified view of how each approach tackles the different problems that appear during the semantic association process. Furthermore, we provide some recommendations on (a) how and when to use each of the approaches according to the characteristics of the data source, and (b) how to improve results by leveraging the strengths of the different approaches.

In this paper, we model knowledge dynamics in agents’ belief bases in a collaborative multi-agent system (MAS). Four change operators are introduced: expansion, contraction, prioritized revision, and non-prioritized revision. For all of them, both constructive definitions and an axiomatic characterization by representation theorems are given. We formally justify minimal change, consistency maintenance, and non-prioritization principles. These operators are based on an epistemic model for multi-source belief revision in which a rational way to weigh the beliefs using a credibility order among agents is developed. The defined operators can be seen as skills added to the agents improving the collective reasoning of a MAS.