Multiagent systems and information retrieval our experience with X.MAS
Highlights
► A multiagent architecture for implementing information retrieval and filtering tasks. ► Personalization issues in information retrieval. ► Six applications that demonstrate effectiveness and usability of the architecture. ► Lesson learnt in applying multiagent systems in the field of information retrieval.
Introduction
Information retrieval (IR) is the task of finding documents of an unstructured nature that satisfies an information need from within large collections (Manning, Raghavan, & Schütze, 2008). IR can cover various and heterogeneous kinds of data and information problems, such as web searching and crawling (Castillo & Baeza-Yates, 2010), personal IR (Thomas, 2005), domain-specific search (Bhavnani, 2002).
The increased availability of documents in digital form, the frequency of their updates and the subsequent information overload require to provide users not simply with “tools”, but also with intelligent systems, such as those based on the multiagent technology. In so doing, users are helped in finding and managing information relevant for their own current interests, profile, and preferences (Klusch, 2001). Different multiagent systems (MAS) have been developed, with different architectures and features, see, for example, (Moukas et al., 1998, Vouros, 2007, Yu et al., 2005, Zhang et al., 2004). In the most cases those systems are populated with at least two types of agents: user agents, which are the interface between a single user and the rest of the system, and information agents, which are connected to information sources and are capable of analyzing and managing it. Moreover, in different application fields, multiagent system can take advantage of their capability to organize themselves and dynamically adapt to changes without explicit external commands (Di Marzo Serugendo et al., 2005, Parunak, 1997).
In this paper, we discuss our experience in building IR systems by resorting to multiagent technology. In particular, we present our results in using X.MAS, a generic multiagent architecture aimed at retrieving, filtering and reorganizing information according to user interests. To this end, we first describe the architecture from both a macro-and a micro-perspective. Moreover, capabilities of X.MAS agents are discussed and the architecture is briefly compared with the main state-of-the-art MAS for IR. To prove the effectiveness of X.MAS, we present six IR systems deployed by customizing X.MAS: (i) NEW.MAS, for creating personalized press reviews; (ii) WIKI.MAS, for classifying Wikipedia contents; (iii) MAM.MAS, for media asset management; (iv) PAA, for recovering plans from the web; (v) SSP.MAS, for predicting secondary structure of proteins; and (vi) SEA.MAS, for monitoring boats in marine reserves.
This paper extends and revises the work by Addis, Armano, and Vargiu (2008a) in which we presented X.MAS. The main extensions are: (i) a more detailed description of X.MAS that takes also into account the improvements introduced in the last two years; (ii) a comparison of X.MAS with state-of-the-art MAS aimed at performing IR tasks; (iii) the description of three further relevant systems developed upon X.MAS; and (iv) for each system, a discussion on the role of the agents and the underlying motivations in adopting a MAS-based approach.
The rest of the paper is organized as follows: in Section 2 we recall the main relevant work on MAS applied in the field of IR. Section 3 presents the generic architecture X.MAS. Section 4 illustrates the six systems devised upon X.MAS. In particular, for each system we first depict the corresponding scenario, then, we present the proposed solution, and, finally, we discuss the underlying motivations in adopting a MAS-based approach. Section 5 ends the paper with some conclusions and a discussion on lesson learnt by using X.MAS for implementing those systems.
Section snippets
Multiagent systems and information retrieval
Autonomous agents and MAS are technologies that have been successfully applied to a number of problems and have been largely used in different application domains (Wooldridge & Jennings, 1995).
As for MAS in IR, in the literature, several centralized agent-based architectures aimed at performing IR tasks have been proposed. Among others, let us recall NewT (Sheth & Maes, 1993), Letizia (Lieberman, 1995), WebWatcher (Armstrong, Freitag, Joachims, & Mitchell, 1995), and SoftBots (Etzioni & Weld,
X.MAS
From a theoretical perspective, an IR task involves three main activities: (i) extracting the required information, (ii) encoding and processing it according to the specific application, and (iii) providing suitable feedback mechanisms to improve the overall performances. Fig. 1 shows an abstract architecture able to perform these activities.
The above problems are typically strongly interdependent in state-of-the-art systems. To better concentrate on these aspects separately, we devised X.MAS,
Personalized press reviews
In this section, we present a MAS explicitly devoted to generate press reviews by (i) extracting articles from Italian online newspapers, (ii) classifying them using text categorization according to user preferences, and (iii) providing suitable feedback mechanisms (Addis, Armano, & Vargiu, 2007). The system has been developed and deployed together with Arcadia Design1 under the project DMC (Digital Media Center) ordered by Cosmic Blue Team.2
Conclusions
In this paper we discussed our experience in using MAS in the IR field. In particular, we presented X.MAS, a generic multiagent architecture explicitly devoted to face to IR tasks. To prove the usefulness and effectiveness of X.MAS, we illustrated six systems built upon it. The implemented systems were aimed at (i) creating personalized press reviews; (ii) classifying Wikipedia contents; (iii) managing media asset; (iv) recovering plans from the web; (v) predicting secondary structure of
References (51)
Information agent technology for the internet: A survey
Data Knowledge Engineering
(2001)- Addis, A., & Borrajo, D. (2010). From unstructured web knowledge to plan descriptions. Information retrieval and mining...
- Addis, A., Armano, G., & Vargiu, E. (2008a). From a generic multiagent architecture to multiagent information retrieval...
- et al.
WIKI.MAS: A multiagent information retrieval system for classifying wikipedia contents
Communications of SIWN
(2008) - Addis, A., Armano, G., Mascia, F., & Vargiu, E., 2007. Hierarchical text categorization through a vertical composition...
- Addis, A., Armano, G., & Vargiu, E. (2007). News retrieval through a multiagent system. In Proceedings of the 7th...
- Addis, A., Armano, G., & Vargiu, E. (2010). Using progressive filtering to deal with information overload. In 7th...
Applications of learning classifier systems
(2004)Why software agents can be effective in biomedical sciences
Multiagent and Grid Systems
(2007)- Armano, G., & Vargiu, E. (2010). A MultiAgent system for monitoring boats in marine reserves. In L. Braubach, J. P....
MASSP3 A system for predicting protein secondary structure
EURASIP Journal of Applied Signal Processing
Digital asset management
Developing multi-agent systems with JADE (Wiley series in agent technology)
Making a strong business case for multiagent technology
Domain-specific search strategies for the effective retrieval of healthcare and shopping information
Collaborative web development: Strategies and best practices for web teams
A weighted nearest neighbor algorithm for learning with symbolic features
Machine Learning
Application of multiple sequence alignment profiles to improve protein secondary structure prediction
Proteins
Self-organization in multi-agent systems
Knowledge Engineering Review
Intelligent agents on the internet: Fact, fiction and forecast
IEEE Expert
Cited by (3)
Multi-agent-based hybrid peer-to-peer system for distributed information retrieval
2023, Journal of Information ScienceMetrIntPair—A Novel Accurate Metric for the Comparison of Two Cooperative Multiagent Systems Intelligence Based on Paired Intelligence Measurements
2018, International Journal of Intelligent SystemsBio-medical literature retrieval from mixed literature bank with the aid of Multi Kernel Fuzzy c-means technique (MK-FCM)
2017, Journal of Theoretical and Applied Information Technology