=Paper=
{{Paper
|id=Vol-1260/paper3
|storemode=property
|title=Social Information Retrieval with Agents
|pdfUrl=https://ceur-ws.org/Vol-1260/paper3.pdf
|volume=Vol-1260
|dblpUrl=https://dblp.org/rec/conf/woa/BergentiPT14
}}
==Social Information Retrieval with Agents==
https://ceur-ws.org/Vol-1260/paper3.pdf
Social Information Retrieval with Agents
Federico Bergenti Agostino Poggi, Michele Tomaiuolo
Dipartimento di Matematica e Informatica Dipartimento di Ingegneria dell'Informazione
Università di Parma Università di Parma
Parma, Italy Parma, Italy
federico.bergenti@unipr.it {agostino.poggi, michele.tomaiuolo}@unipr.it
Abstract—With the widespread adoption of online social incorporating individuals into the model, IR techniques gain a
networks as a crucial means for communication, social greater insight into the documents under observation. New
information retrieval is becoming one of the most interesting associations between entities become apparent, e.g., individuals
areas of research in terms of the large number of–theoretical and appear in their role as information producers or information
practical–issues that it encompasses. We argue that agent consumers, queries relate to an individual’s information needs,
technology is central in supporting the decentralization of next and they describe a topic that falls into the interests of an
generation online social networks and the synergistic pairing of individual.
agents and social networks is evident, if nothing else, because
members of a social network interact as agents do in a multi- The ultimate motivation for social IR is rooted in the belief
agent system. In this paper we investigate the possibilities that that an information producer and his/her product cannot be
agent technology can offer to social information retrieval and we separated, and, likewise, information and its consumers cannot
emphasize the role that agents and multi-agent systems can play be separated.
by presenting Blogracy, an agent-based online social network Understanding the social context in which the production
system. and consumption of information takes place is especially
important when only limited understanding of the information
Social information retrieval, online social networks, multi-agent under consideration is available. Traditional IR techniques are
systems based solely on analysing the content of documents and, while
very successful in many contexts, they fail when the
I. INTRODUCTION information of documents under observation is partial. In this
sense, social IR can be understood as a formalization of the
Nowadays it is common opinion that most Information search techniques that we commonly use to assess the quality
Retrieval (IR) systems and related tasks are more than ever of information–by looking at the author’s standing in his/her
embedded in rich contexts. Documents no longer exist on their community. The same principle can be applied to other
own: (i) they are connected to other documents; (ii) they are instances of information production and consumption in a
associated with the individuals that contributed to produce social environment: we tend to judge information also on the
them, and with the individuals that, possibly partially, consume basis of the reputation that its producers and respective
them; (iii) they are dependent of the social networks of their consumers have in their social context.
respective producers and consumers; and (iv) they are related
to the context in which individuals operate. All such features For these, and possibly many other reasons, we believe that
enrich documents and the correct use of them can drastically social IR poses new challenges and questions that are worth
increase the performance of IR techniques. investigating. We also believe that the quality of contextual
information available to social IR techniques heavily depends
Social IR can be broadly defined as the synergistic on the tools that the social IR system can adopt to grasp social
embedding of information about social networks of individuals aspects relevant to IR tasks.
and their relationships into IR processes (see, e.g.,
[10][12][26]). Individuals and their social networks are cornerstones of
social IR, but users and online social networks can perform
The traditional models for IR have to do with documents, orders of magnitude better. Social networks are typically
queries, and their relations. For example, a document is described as finite sets of actors and relations defined on them
relevant to a query, but a document may reference to other (see, e.g., [40]). In this context, an actor is essentially any
documents and, likewise, a query may be related to another social entity, such as an individual, a corporate, or a collective
query. In a similar spirit, social networks model individuals and social unit; and a relationship can be any kind of social tie that
their relations, like friends and family, acquaintances, and establishes a link between a pair of actors. Nowadays, the most
collaborators (see, e.g., [4][10][40]). Unfortunately, traditional widely known social networks are Web platforms, often called
IR techniques do not model individuals, neither in their role as online social networks, where users not only put or read
users of the system, nor as authors of the retrieved documents. content, but they are also linked with relationships. The
This circumstance severely limits the contextual information diffusion of online social networks is opening new scenarios
available to the IR techniques, and the promise of social IR is for envisaging novel kinds of applications, either to support
to boost the performance of IR techniques by means of the new social networking activities, or to exploit established
integration of socially relevant contextual information. By
relationships among users and use them to offer higher-level e.g., having a given capability or expertise. This is quite similar
services. to the problem of navigating one’s social network in search for
someone with a given expertise or for an answer to a specific
With this in mind, we believe that online social networks question. In an enterprise setting, this is the problem of looking
are still not sufficient because in their current incarnation they inside the organization for someone able to solve a specific
tend to be highly centralized and to form, sometimes huge, problem or able to answer to a specific question. When solved
islands. The recent clamor about the PRISM program and the with agent-based techniques, this problem resembles the
release of classified documents by Edward Snowden [19] has collaborative filtering one and is usually termed as expert
also raised many questions about the privacy issues of current finding, and authors use such definitions interchangeably.
social networking applications. We think that social IR can be
taken to its full potential by eliminating the boundaries of The expert finding problem is similar to Milgram’s original
current online social networks and by fostering IR tasks that problem in that the social network of each node is the search
may break across networks. We believe that agent technology space in which the request is processed. It should be
is crucial to enable such an envisioned decentralization of emphasized that both problems strongly rely on the local
online social networks because of the inherent decentralized search ability and the occurrence of the small world
nature of multi-agent systems (see, e.g., [9][29]) and because phenomenon, i.e., on the fact that two random individuals are
of their intrinsic characteristics in terms of management of preferably mostly connected by short chains of
trust, privacy, and reputation (see, e.g., [2][7][36][37][39]). acquaintanceships. If social networks were not searchable it
would be impossible to efficiently find a person matching some
In the following section we outline the major features that criteria unless personally known and, then, the Milgram’s
the synergistic pairing of online social networks and agent experiment would have failed. On the other hand, if the chains
technology offers, and we survey recent research effort that were very long, the search would be not feasible.
explored such a combination in various contexts. Finally, we
present an agent-based online social network system, namely A pioneering research on this subject was done in [24][25].
Blogracy [15][16], that promotes decentralization and that is These papers describe ReferralWeb, an agent based interactive
therefore a solid base for taking social IR to its full potential. system for reconstructing, visualizing, and searching social
networks on the Web whose main focus is selecting an expert
II. AGENT TECHNOLOGY AND ONLINE SOCIAL NETWORKS of a given field in one’s (extended) social network.
In order to understand the relationship between multi-agent In ReferralWeb a social network is modelled by a graph
systems and social networks it is important to understand the where the nodes represent individuals and an edge between
intrinsic computational properties of social networks. The first nodes indicates that a direct relationship between the
insights on such properties came from Milgram’s experiment individuals has been discovered. For ReferralWeb a direct
that led to the investigation of the so-called small world relationship is implied when the names are in close proximity
phenomenon [28]. In Milgram’s experiment, a group of in any document publicly available on the Web, e.g., home
randomly chosen people received the name and address of pages, co-authorship in published papers, or organization charts
another randomly chosen person living in a distant city. Then, in institutional Web sites. ReferralWeb does not require its
people were asked to route a mail message toward the target users to fill a user profile describing their skills.
person chosen only among their friends or close acquaintances. The constructed network is then used to guide the search
The experiment pointed out that: (i) people are connected for people or documents in response to user queries. A person
through very short chains of acquaintances, with a 5-6 links can: (i) ask to find the chain between himself/herself and a
length, in average; and (ii) people is able to route the messages named individual; (ii) search for an expert in a given topic
to the target person using local information and performing providing a maximum social radius (the number of links in the
local actions. chain connecting the person performing the query with the
A result of the Milgram’s experiment is that the behavior of expert); and (iii) request a list of documents written by people
people was similar to that of rational autonomous agents. In close to a given expert.
fact, every person choses his/her successor in his/her list of The key idea of ReferralWeb is to use the social network to
acquaintances considering elements like geographical make more focused and effective searches. It is not meant to be
proximity or profession similarity, which is essentially using a tool to create social networks, i.e., to help people socializing.
only local and elementary information to pursue a global ReferralWeb also emphasizes the importance of the referral
complex goal, with no need to use their humanity. From our chains themselves as means to build trust on the selected
point of view, this is a particularly relevant conclusion, since it experts.
points to the emergence of a global behavior from local
strategies, a feature that is one of the key properties of multi- MARS is a multi-agent referral system that finds experts on
agent systems. the basis of personal agents able to learns the user’s
preferences and interests, and able to build an expertise model
More recently, the studies on the small world problem led of the other users on the basis of their responses [42]. Each
to two computationally-based approaches to search for people user is assigned an agent who: (i) learns the user’s preferences
within social networks (a comprehensive review of different and interests, and (ii) maintains a view of its user’s
algorithms and their performance is presented in [1]). The acquaintances, that are used to prioritize incoming queries,
original experiment of Milgram led to a machine-based possibly issuing referrals when other users might be more
approach consisting in the problem of looking for a remote suitable to answer a given query. Each agent first rates,
agent given its unique identifier. A successive approach deals according to the user’s feedback, those agents that provided an
with finding a specific agent who matches a given criterion, answer and those agents that referred to them and, then it
modifies its neighbors accordingly. Consequently, the referral Shine provides a personal agent to each and every single
system evolves to reflect the changes in the social network. user and three core modules compose each agent: the person
database, the plan execution module and the communication
A response to a query specifying what information is being module. In addition, one or more applications are installed in
sought, if given, may consist of an answer or a referral, each agent. Such applications provide their services to the user
depending on the query and on the expertise of the answering by means of functionalities of the core modules via a dedicated
agent. If an agent is reasonably confident that its expertise API.
matches the query, it directly answers; otherwise, it yields
referrals to other supposedly expert agents. The person database of Shine holds data on people and on
personal agents. The data include information on the agent and
Each agent maintains models of its acquaintances. An agent on the user whom the agent is associated with, as well as other
sends its query initially only to some of its neighbors, that are agents and people known to the agent. An agent holds the data
the individuals with the closest acquaintances. The agent who required to form a community that is suitable to the user in the
receives a referral may pursue it even if the referred party is not person database and it exchanges data among other agents
already an acquaintance; good acquaintances are going to be when necessary. In the Shine architecture, the user and his/her
promoted to neighbors on an intuitive basis. When new personal agent correspond in a one-to-one manner. Therefore,
neighbors are considered, some of previous ones will be in the person database, data on both a user and his/her personal
discarded, since the number of neighbors is bounded. The agent are stored without distinguishing between them.
authors of MARS decided that reputation should increase
slowly, but it should fall out quickly, and that rewards and In order to support communities, Shine’s authors added the
penalties are greater for agents nearer to the answering agent. concept of person set. Each community is represented in the
This implies that a bad decision results in bad reputation, but if person database as a person set and the framework provides
agents just started a chain of referrals leading to a bad agent, operations for dealing with such sets, e.g., functions to
then the penalty is modest. broadcast messages to the members of a community. In this
way Shine agents can flexibly determine the range of
The expertise model is captured through a classical vector broadcasting by regarding a person set as the destination list.
space model [38]. Term vectors are used to express both the
profile of the user and the acquaintance model for each of its In Shine a peer-to-peer network is formed directly
acquaintances. Since a term vector also models the required connecting the communication modules of groups of agents.
expertise, the cosine of the angle between the user vectors with The function of such modules is simply to exchange messages
the subject vector yields the competence of a user in a given with each other. Given the fact that the agents live in a
subject. Intuitively, when there are two agents with expertise in ubiquitous computation environment, the module is layered so
the same direction, the one with the greater expertise is more that only the lower layer depends on the environmental details.
desirable.
Agents in Shine are goal-driven through plans: a plan is
Each agent learns its user’s profile and its acquaintance description of agent action rules. Multiple plans are executed
models based on an evaluation of the received answers as well concurrently in the plan execution module of each agent. Some
as on the referrals that led to them. A referral graph, which is plans are prepared to perform services of applications while
local to each agent, encodes how the computation spreads, as a other plans are provided by Shine to do fundamental or
query originates from an agent, and referrals or answers are common tasks. A plan acts in response to external events, e.g.,
sent back to this agent. receiving a message from another agent, a user input or a
modifications in the person database.
Yenta is a matchmaking system that helps people with
similar interests to get in touch [14]. Yenta agents do not query SNIS is a multi-agent system where agents utilize the
the Web; instead, they scan user’s e-mails, Usenet posts and connections of a user in the social network to facilitate the
(possibly) documents in order to discover their users’ interests search for items of interest [21]. In particular, each agent is
and hobbies. The idea is that many potentially interesting associated with a user and it observes the user’s activities and,
people do not write publicly and so they become invisible to in particular, the ratings and comments provided by the user to
tools relying on public data. Collected data are then used to items retrieved from the social network. SNIS has been
introduce users’ to each other. Considering that in the 90’s Web experimented in the Flickr domain [27]; the system scans
communities were built around the idea of common interests photos posted by all of the user’s contacts and gathers statistics
rather than on personal acquaintance, the system was a truly about their categories and user comments (which represent user
distributed social networking system, at least for the time. interest) and such information is used to facilitate the search for
items of interest.
Shine (SHared INternet Environment) [41] is a fully peer-
to-peer framework for network community support. The
system has been implemented and a presented in [41]. The III. BLOGRACY
framework provides design guidelines and enables different It is common opinion that multi-agent systems can play an
applications to share program components and to cooperate, important role to support completely decentralized or federated
and it features a peer-to-peer architecture through which social networking platforms. Indeed, one of the very specific
personal agents can flexibly form communities where users can features of multi-agent systems is the sociality of agents, i.e.,
exchange information with peer agents. Essentially, Shine is a their ability to communicate in a semantic way (see, e.g.,
middleware for collaborative workspaces especially tailored to [6][32]) and to develop trust relationships among them.
implement various collaborative workspaces. Moreover, agents can express their communication acts by
means of acknowledged standards for interoperability among
diverse systems, like FIPA [13], and they can exchange
Figure 1. The multi-agent architecture of Blogracy.
messages directly in a peer-to-peer way. Therefore, it is not network, to make new acquaintances with users with common
surprising that these two technologies are often applied interests, to find interesting content hidden in less relevant data
together for developing advanced social platforms. or from new sources.
In particular, multi-agent systems have been used as: (i) an Both kinds of features of agents and multi-agent systems
underlying layer, or middleware, for developing social have been already integrated in the design of Blogracy
networking platforms; and (ii) a technology to increase the [7][15][33], an agent-based system whose goal is to provide
autonomous and intelligent behavior of existing systems. users with adaptive and composite services on top of core
features. At the lower level, Blogracy uses widespread and
For the first type of applications of multi-agent systems, stable peer-to-peer technologies, such as distributed hash tables
many of the distinguishing features of multi-agent systems can and the BitTorrent protocol, for coping with the intrinsic
be fully exploited. Indeed, multi-agent systems provide defects of centralized architectures and to become the basis of
semantic communication among agents, which is handy for solid distributed social networking platforms. At the higher
expressing all the different actions that users can perform in a level, it takes advantage of multi-agent systems for simplifying
social platform. The different types of messages can be the implementation of social network services in a
understood according to their meaning and applied according decentralized setting.
to existing trust relations among the users and their respective
agents. In addition, complex negotiation protocols can help The architecture of Blogracy is modular and composed of
creating acknowledgements and trust among users, in an two basic components: (i) an underlying module for basic file
automatic or assisted way, without exposing sensitive data. sharing and DHT operations, built as an extension of existing
Mobility can also be useful for moving the computation closer implementations, and (ii) an OpenSocial container, i.e., a
to data, if massive analysis has to be performed, but it can also module providing the services of the social platform to the
be handy for adding functionality to a node of a decentralized local user through a Web interface. Additionally, Blogracy
social platform or to a user’s client application. supports autonomous agents to provide recommendations of
both users and content, personalization of results, and trust
In the second type of applications of multi-agent systems, negotiation mechanisms.
agents are mainly used because of their proactive and reactive
behaviors that can provide recommendations of both users and Blogracy relies only on users’ nodes for its operation.
content, and that can enable the personalization of results. Therefore, users need to perform background tasks on their
Reactive abilities are particularly important in a social own, in a distributed and decentralized way.
networking environment where interesting events happen
frequently and where users can be easily distracted by the huge A layer of agents takes charge of assisting the user in
information flow, which is associated with highly connected finding new interesting content and connections, and in
social networks. Sensing the environment and executing pushing the local user’s activities to followers.
automatic tasks can reduce this overload significantly. Goal- Figure 1 sketches the multi-agent architecture of Blogracy.
oriented behaviors, on the other hand, can support users in A Personal Assistant (PA) monitors the local user’s actions in
prosecuting their long term objectives about friend and content the platform and it learns the user’s profile, beyond information
discovery, i.e., to discover known persons registered to the provided explicitly. The PA receives the user’s queries,
forwards them to the available Information Finders (IF) and it that are subset of other groups (or of the set union of all groups,
presents the results to the user. Moreover, a PA provides the i.e., only “friends” are part of a proximity group). The NM
local user with recommendations about possibly interesting agent informs the OS agent when users enter and leave the
content and connections available in the network. Another task proximity group and the latter notifies the OpenSocial
performed by the PA is the personalization of results. Indeed, container about it.
as a social network becomes larger and more richly
interconnected, users unavoidably face some form of On the other hand, a location group is associated with the
information overflow. A PA, using a user’s profile, can arrange users in the proximity of a given location (e.g., a classroom or
presented data in a way that highlight the most interesting a museum room) and it has a host (i.e., a node) that both
pieces of information. identifies and supports the group. Moreover, a location group is
associated with a location profile maintained either on the
An IF is an agent that searches information on the central server or on its host. In fact, a location, although
repository contained in the node where it lives, through an logically different from a regular user, works in the same way
automatic TF-IDF indexing algorithm and explicit hashtags and a location group is essentially a proximity group for the
associated with local posts. It provides such information both location.
to its user and to other trusted users. An IF receives users’
queries, finds appropriate results and filters them by using its The availability of a generic TN agent is also important
user’s access policies. An Information Pusher (IP) is an agent since users joining a proximity group or a location group are
that monitors the changes in the local repository and that not necessarily connected a priori in the social network, and
pushes new information to the PA of interested subscribers that they may need to acknowledge their profile attributes before
are currently connected. An IP can forward content produced practical social interaction. Such a negotiation requires the
by the local user and by her/his remote acquaintances to other controlled exchange of credentials and of policies, without
contacts, according to privacy preserving policies and to recent disclosing unnecessary sensible information, yet establishing
queries made by other users. trust if possible [2][36][37][39]. In [7], a generic library
supporting zero-knowledge proof for attribute verification is
Over the OpenSocial container, Blogracy can also provide presented. The same mechanisms can also facilitate the
functionalities for pervasive online social networking, creation of trust in social networks.
specifically for creating locality and proximity groups. In this
case, the system has to rely on highly adaptive services both to Agents present different degrees of autonomy and
sustain the basic operations of the location-based social intelligence. For example, lower level agents are mostly
networking and to provide advanced functionalities. For this reactive; e.g., they inform the NM agent when a new node is
purpose, each node of the social network has to host multiple discovered. The NM agent itself has some degrees of autonomy
agents, with different levels of agency [31][34][35]. Some of and intelligence, and it has the following duties: (i) it
the more important agents are: aggregates information from the agents that discover new
peers; (ii) it informs the OS agent of the state of neighborhood;
• The Neighborhood Manager (NM) agent, which (iii) it tries to present a consistent view, merging data from the
cooperates with lower level agents to discover the different sources; and (iv) it configures the discovering agents
users in its neighborhood; according to high-level criteria, such as battery consumption
and hardware availability.
• The Trust Negotiator (TN) agent, which is involved in
the decisions regarding privacy and access rules; and The OS agent is basically a gateway to the OpenSocial
container and it translates the other agents’ requests for the
• The OpenSocial (OS) agent, which provides a bridge OpenSocial container. A TN agent is a true agent that performs
towards the underlying modules of Blogracy. potentially complex negotiations on its user’s behalf and,
depending on the configuration, it may work in full autonomy.
A user may own multiple nodes (e.g., an instance on the
smartphone and an instance on his home computer) and, since
the actual location of the user is important for our application, IV. CONCLUSIONS
the nodes in the different devices negotiate which one should This paper outlines a very promising line of research: the
be considered active (i.e., which one determines the user use of the entire spectrum of agent technology to provide
location). In fact, the nodes can either determine the device that concrete support to innovative social IR tasks. Agents and
registered an explicit user action or they can ask to the user to multi-agent systems naturally models social networks, and they
select the device he/she is currently using. can even implement large-scale online social networks as
Apart from the personal circles defined by each user, we nowadays agent technology is considered a mature tool capable
also have two additional kinds of groups: proximity groups and of supporting mission-critical, large-scale software systems
location groups. Proximity groups are centered on each (see, e.g., [3][5][8][17][22][23][30]).
member of the social networking system and they represent Moreover, the peculiar management of decentralization and
physical closeness to such a member. Proximity groups are the sophisticated treatment of privacy and reputation issues
extremely fluid in the sense that users can physically move and make agents and multi-agent systems ideal tools to provide
consequently the set of users belonging to a proximity group insightful contextual information to social IR techniques. In
varies over time. Each user configures the hysteresis, or sticky- particular, Blogracy breaks the traditional centralized approach
ness, of his proximity group, i.e., how long other users are to the implementation of online social networks and it opens to
considered part of it after they are no longer physically close to new sources of contextual information that can be obtained by
him/her. Although a proximity group may be entirely public, observing documents and individuals across multiple social
for privacy reasons it is safer to consider only proximity groups networks. All in all, the decentralization that agent technology
ensures define a novel features of documents, individuals, and [19] M. Greene, “Where Has Privacy Gone? How Surveillance Programs
relations: how they spread across different online social Threaten Expectations Of Privacy”. The John Marshall Journal of
Information Technology & Privacy Law, 30(4), 5. 2014.
networks, and how they change in such a spreading over time.
Moreover, the simple fact of observing individuals and [20] D. Greenwood, and M. Calisti, “Engineering Web service-agent
integration,” in IEEE International Conference on Systems, Man and
documents overlapping different online social networks is Cybernetics (Vol. 2), The Hague, Netherlands, 2004, pp. 1918-1925.
immediately usable as a relevant source of contextual
[21] A. Gursel, and S. Sen, “Improving Search In Social Networks by Agent
information. Based Mining,” in 21st International Joint Conference on Artifical
Intelligence, San Francisco, CA, USA, 2009, pp. 2034-2039.
In conclusion, we argue that agent technology provides
solid and mature tools to support the design and [22] M.N. Huhns et al., “Research Directions for Service-Oriented
Multiagent Systems,” IEEE Internet Computing, vol. 9, no. 6, pp. 65-70,
implementation of novel social IR tools, and we believe that no 2005.
effective social IR can take place if it would restrict to a single,
[23] N. Jennings, J. Corera, and I. Laresgoiti, “Developing industrial multi-
even if enormous, online social network. agent systems,” in First International Conference on Multi-Agent
Systems (ICMAS-95), 1995, pp. 423–430.
REFERENCES [24] H. Kautz, B. Selman, and M. Shah, “Combining Social Networks and
Collaborative Filtering,” Communications of the ACM, vol. 40 no. 3, pp.
[1] L. Adamic, and E. Adar, “How to search a social network,” Social
63-65, 1997.
Networks, vol. 27, no. 3, pp. 187-203, 2005.
[25] H. Kautz, B. Selman, and M. Shah, “The Hidden Web,” AI Magazine,
[2] F. Agazzi, and M. Tomaiuolo. “Trust Negotiation for Automated Service
vol. 18, no. 2, pp. 27-36, 1997.
Integration”, in CEUR Workshop Proceedings 1099, WOA 2013.
[26] S.M. Kirsch, M. Gnasa, and Armin B. Cremers. “Beyond the Web:
[3] F. Bellifemine, G. Caire, A. Poggi, and G. Rimassa, “JADE: a Software
Retrieval in social information spaces”. Procs. 28th European Conf. on
Framework for Developing Multi-Agent Applications. Lessons
Advances in Information Retrieval, 84-95, Springer-Verlag, 2006.
Learned,” Information and Software Technology Journal, vol. 50, pp.
10-21, 2008. [27] K. Lerman, A. Plangrasopchok, and C. Wong, “Personalizing results of
image search on Flickr,” in AAAI workshop on Intelligent Techniques
[4] F. Bergenti, E. Franchi, and A. Poggi, “Selected Models for Agent-based
for Web Personlization, Vancouver, Canada, 2007, pp. 65-75.
Simulation of Social Networks,” in Social Networks and Multi Agent
Systems Symposium (SNAMAS 2011), pp. 27-32. 2011. [28] S. Milgram, “The small world problem,” Psychology today, vol. 1, no. 1,
pp. 61-67, 1967.
[5] F. Bergenti, and A. Poggi, “An Agent-Based Approach to Manage
Negotiation Protocols in Flexible CSCW Systems,” in 4th Int. Conf. on [29] J. Muller, “Architectures and applications of intelligent agents: A
Autonomous Agents, pp. 267-268. 2000. survey,” Knowledge Engineering Rev., vol. 13, no. 4, pp. 353-380, 1998.
[6] F. Bergenti, A. Poggi, M. Tomaiuolo, and P. Turci, “An Ontology [30] A. Negri, A. Poggi, M. Tomaiuolo, and P. Turci, “Dynamic Grid Tasks
Support for Semantic Aware Agents”. Lecture Notes in Computer Composition and Distribution through Agents,” Concurrency and
Science, 3529/2006, pp. 140-153. Springer. 2006. Computation: Practice and Experience, vol. 18, no. 8, pp. 875-885,
2006.
[7] F. Bergenti, L. Rossi, and M. Tomaiuolo, “Towards Automated Trust
Negotiation in MAS,” in WOA 2009. Parma, Italy. 2009. [31] A. Negri, A. Poggi, M. Tomaiuolo, and P. Turci, “Agents for e-Business
Applications”, in: 5th Int. Joint Conf. on Autonomous Agents and Multi-
[8] F. Bergenti, A., Poggi, and M. Somacher, “A collaborative platform for
Agent Systems (AAMAS-2006), pp. 907-914. ACM. 2006.
fixed and mobile networks,” Communications of the ACM, vol. 45, no.
11, pp. 39-44, 2002. [32] A. Poggi, “Developing ontology based applications with O3L,” WSEAS
Trans. on Computers, vol. 8, no. 8, pp. 1286-1295, 2009.
[9] R. Bordini, M. Dastani, J. Dix, and A. Fallah-Seghrouchni, “Multi-
Agent Programming: Languages, Platforms and Applications”. [33] A. Poggi, and M. Tomaiuolo. “A DHT-Based Multi-Agent System for
Multiagent Systems, Artificial Societies, and Simulated Organizations, Semantic Information Sharing”, in Studies in Computational
vol. 15. Berlin, Germany: Springer, 2005. Intelligence, 439/2013, pp. 197-213. Springer, 2013.
[10] K. Chard, S. Caton, O. Rana, and K. Bubendorfer, “Social cloud: Cloud [34] A. Poggi, M. Tomaiuolo, and P. Turci, “An Agent-Based Service
computing in social networks,” in 2010 IEEE 3rd International Oriented Architecture”. 157-165, In: WOA 2007, Genova, Italy.
Conference on Cloud Computing, Miami, FL, USA, 2010, pp. 99-106. [35] A. Poggi, M. Tomaiuolo, and P. Turci, “Extending JADE for agent grid
[11] W. Croft, D. Bruce, T. Metzler, and T. Strohman, “Search engines: applications”, in Proc. WET ICE 2004, pp. 352-357. IEEE. 2004.
Information retrieval in practice”. Addison-Wesley, 2010. [36] A. Poggi, M. Tomaiuolo, and G. Vitaglione, “A security infrastructure
[12] D. Horowitz, and D. K. Sepandar. “The anatomy of a large-scale social for trust management in multi-agent systems”, in Trusting Agents for
search engine,” in Procs. 19th Int. Conf. on WWW, 2010, pp. 431-440. Trusting Electronic Societies, pp. 162-179. Springer. 2005.
[13] FIPA (2014, February 10). “FIPA Specifications” [Online]. Available at [37] A. Poggi, M. Tomaiuolo, and G. Vitaglione, “Security and trust in agent-
http://www.fipa.org/. oriented middleware”, in On The Move to Meaningful Internet Systems
(OTM 2003), pp. 989-1003. Springer. 2003.
[14] L. Foner, “Yenta: a multi-agent, referral-based matchmaking system,” in
First International Conference on Autonomous Agents, Marina del Rey, [38] G. Salton, and M. McGill, “Introduction to modern information
CA, USA, 1997, pp. 301-307. retrieval”. New York, NY, USA: McGraw Hill. 1983.
[15] E. Franchi, A. Poggi and M. Tomaiuolo. Open Social Networking for [39] M. Tomaiuolo, “dDelega: Trust Management for Web Services”.
Online Collaboration. International Journal of e-Collaboration 9(3), International Journal of Information Security and Privacy (IJISP), 7(3),
2013. 53-67, 2013. ISSN:1930-1650. doi:10.4018/jisp.2013070104.
[16] E. Franchi, and M. Tomaiuolo, “Distributed Social Platforms for [40] S. Wasserman, and K. Faust, “Social network analysis: methods and
Confidentiality and Resilience”. In: Social Network Engineering for applications”. Cambridge, UK: Cambridge University Press. 1994.
Secure Web Data and Services, IGI Global, 2013. [41] S. Yoshida, K. Kamei, T. Ohguro, and K. Kuwabara, “Shine: a peer-to-
[17] F. Gandon, A., Poggi, G., Rimassa, and P. Turci, “Multi-Agents peer based framework of network community support systems,”
Corporate Memory Management System,” Applied Artificial Computer Communications, vol. 26 no. 11, pp. 1199-1209, 2003.
Intelligence Journal, vol. 16 no. 9-10, pp. 699-720, 2002. [42] B. Yu, and M Singh, “Searching social networks,” in Second
[18] M. Genesereth, and S. Ketchpel “Software Agents,” Communications of international joint conference on Autonomous agents and multiagent
the ACM, vol. 37, no 7, pp. 47-53, 1994. systems, NY, USA, 2003, pp. 65-72.