Holger Billhardt

A context vector model for information retrieval

In the vector space model for information retrieval, term vectors are pair-wise orthogonal, that is, terms are assumed to be independent. It is well known that this assumption is too restrictive. In this article, we present our work on an indexing and retrieval method that, based on the vector space model, incorporates term dependencies and thus obtains semantically richer representations of documents. First, we generate term context vectors based on the co-occurrence of terms in the same documents. These vectors are used to calculate context vectors for documents. We present different techniques for estimating the dependencies among terms. We also define term weights that can be employed in the model. Experimental results on four text collections (MED, CRANFIELD, CISI, and CACM) show that the incorporation of term dependencies in the retrieval process performs statistically significantly better than the classical vector space model with IDF weights. We also show that the degree of semantic matching versus direct word matching that performs best varies on the four collections. We conclude that the model performs well for certain types of queries and, generally, for information tasks with high recall requirements. Therefore, we propose the use of the context vector model in combination with other, direct word-matching methods.

Organising MAS: a formal model based on organisational mechanisms

In this paper, we propose a general formal framework for organising multiagent systems whose participants are rational agents. This model is based on the idea of organisational mechanisms. These are mechanisms introduced in a multiagent system with the aim of influencing the behaviour of the agents towards more effectiveness with regard to some objectives. We define two kinds of organisational mechanisms: i) informative mechanisms which provide additional information to agents, that may persuade agents to behave in a certain way, and ii) regulative mechanisms which produce changes in the environment of the agents, that may impose certain behaviours. We also define some properties of these mechanisms which will make it possible to prove certain characteristics of organised multiagent systems. Finally, we present a discussion about how the social concepts proposed by different organisational paradigms can be considered as either informative or regulative organisational mechanisms.

Integrating Trust in Virtual Organisations

Organisational models cannot only be used to structure multiagent systems but also to express behaviour constraints for agents in open environments. However, sometimes these behaviour constraints cannot be exhaustively enforced, and some agents may transgress the norms put forward by a Virtual Organisation. This poses an additional burden on agents, as they cannot be sure that their acquaintances will behave as prescribed. Trust and reputation mechanisms are of particular relevance to this respect, as they are commonly used to infer expectations of future behaviour from past interactions. In this paper we argue that, on the one hand, the a priori structure of Virtual Organisations can be useful to improve the efficiency of trust and reputation mechanisms, and that, on the other hand, such mechanisms provide relevant information for agents that are part of Virtual Organisations. For this purpose, we identify relevant aspects of existing organisational (meta-)models, and outline a reputation mechanism for Virtual Organisations that integrates these aspects. The dynamics of this mechanism is illustrated by an example.

Service discovery acceleration with hierarchical clustering

This paper presents an efficient Web Service Discovery approach based on hierarchical clustering. Conventional web service discovery approaches usually organize the service repository in a list manner, therefore service matchmaking is performed with linear complexity. In this work, services in a repository are clustered using hierarchical clustering algorithms with a distance measure from an attached matchmaker. Service discovery is then performed over the resulting dendrogram (binary tree). In comparison with conventional approaches that mostly perform exhaustive search, we show that service-clustering method brings a dramatic improvement on time complexity with an acceptable loss in precision.

An Adaptive Sanctioning Mechanism for Open Multi-agent Systems Regulated by Norms

This work presents an adaptive sanctioning mechanism that can be applied in open multi-agent systems that are regulated through norms. This mechanism tries to identify the attributes of the environment that have some influence on agents decision making and uses such attributes to define sanctions that may prevent norm violations. Our approach adapts sanctions to particular agents and particular environmental states. The mechanism is deployed by using an infrastructure of institutional agents in charge of adapting and applying sanctions to external agents. The proposal is evaluated empirically in a p2p scenario, comparing it with a traditional sanctioning approach for normative systems.

Effective use of organisational abstractions for confidence models

Trust and reputation mechanisms are commonly used to infer expectations of future behaviour from past interactions. They are of particular relevance when agents have to choose appropriate counterparts for their interactions as it may also happen within virtual organisations. However, when agents join an organisation, information about past interactions is usually not available. The use of organisational structures can tackle this problem and can improve the efficiency of trust and reputation mechanisms by endowing agents with some extra information to choose the best agents to interact with. In this context, we present how certain structural properties of virtual organisations can be used to build an efficient trust model in a local way. Furthermore, we introduce a testbed (TOAST) that allows to analyse different trust and reputation models in situations where agents act within virtual organisations. We experimentally evaluate our approach and show its validity.

Supporting Medical Emergencies by MAS

In the emerging field of m-Health, advanced applications provide healthcare to people anywhere, anytime using broadband and wireless communications, as well as mobile computing devices. The notion of Service-Oriented Multi-Agent Systems (SOMAS) that has recently been proposed appears to adequately capture the requirements of applications in this field. The THOMAS abstract architecture and software platform supports the construction of SOMAS around an organisational metaphor. In this paper we present an application prototype built on top of THOMAS for a real-world mHealth scenario: mobile medical emergency management in an urban area.

Learning retrieval expert combinations with genetic algorithms

The goal of information retrieval (IR) is to provide models and systems that help users to identify the relevant documents to their information needs. Extensive research has been carried out to develop retrieval methods that solve this goal. These IR techniques range from purely syntax-based, considering only frequencies of words, to more semantics-aware approaches. However, it seems clear that there is no single method that works equally well on all collections and for all queries. Prior work suggests that combining the evidence from multiple retrieval experts can achieve significant improvements in retrieval effectiveness. A common problem of expert combination approaches is the selection of both the experts to be combined and the combination function. In most studies the experts are selected from a rather small set of candidates using some heuristics. Thus, only a reduced number of possible combinations is considered and other possibly better solutions are left out. In this paper we propose the use of genetic algorithms to find a suboptimal combination of experts for a document collection at hand. Our approach automatically determines both the experts to be combined and the parameters of the combination function. Because we learn this combination for each specific document collection, this approach allows us to automatically adjust the IR system to specific user needs. To learn retrieval strategies that generalize well on new queries we propose a fitness function that is based on the statistical significance of the average precision obtained on a set of training queries. We test and evaluate the approach on four classical text collections. The results show that the learned combination strategies perform better than any of the individual methods and that genetic algorithms provide a viable method to learn expert combinations. The experiments also evaluate the use of a semantic indexing approach, the context vector model, in combination with classical word matching techniques.

Dynamic Coordination in Fleet Management Systems: Toward Smart Cyber Fleets

Fleet management systems are commonly used to coordinate mobility and delivery services in a broad variety of domains. However, their traditional top-down control architecture becomes a bottleneck in open and dynamic environments, where scalability, proactiveness, and autonomy are becoming key factors for their success. Here, the authors present an abstract event-based architecture for fleet management systems that supports tailoring dynamic control regimes for coordinating fleet vehicles, and illustrate it for the case of medical emergency management. Then, they go one step ahead in the transition toward automatic or driverless fleets, by conceiving fleet management systems in terms of cyber-physical systems, and putting forward the notion of cyber fleets.

Using genetic algorithms to find suboptimal retrieval expert combinations

A common problem of expert combination approaches in Information Retrieval (IR) is the selection of both, the experts to be combined and the combination function. In most studies the experts are selected from a rather small set of candidates using some heuristics. Thus, only a reduced number of possible combinations is considered and other possibly better solutions are left out. In this paper we propose the use of genetic algorithms to find a suboptimal combination of experts for a document collection. Our system automatically determines both, the experts to be combined and the parameters of the combination function. We test and evaluate the approach on four classical text collections. The results show that the learnt combination strategies perform better than any of the individual methods and that genetic algorithms provide a viable method to learn expert combinations.