Search

Seeking support from their environment is important for people suffering from a depression. People usually have different social networks to which they are attached with different ties. In this paper, a computational model is presented that describes the selection of network members for seeking support based on the strength of the tie to people in the network and personal characteristics. The model has been implemented in a simulation environment. Simulations of different scenarios show that specific personality traits and environmental settings indeed lead to a pattern of social disengagement or a preference for strong or weak tie support. A mathematical analysis proofs that such equilibria are indeed a consequence of the model. © 2009 IEEE.

This paper is focusing on exact Bayesian reasoning in systems of agents, which represent weakly coupled processing modules supporting collaborative inference through message passing. By using the theory on factor graphs and cluster graphs we (i) analyze the suitability of the existing approaches to modular inference with respect to a relevant class of domains and (ii) derive methods for construction of modular systems, which support globally coherent Bayesian inference without compilation of secondary probabilistic structures spanning multiple modules. In the proposed approach dependencies between inference modules are reduced through targeted instantiation of variables, which is based on the analysis of cluster graphs.

A Multi-Agent System is often conceived as an organization of autonomous software agents that participate into social and evolving structures (e.g., organizational configurations) suitable to deal with highly dynamic environments. Nevertheless, systems based on agent technologies rarely capitalize on their potentials since their systemic properties— e.g., flexibility, robustness and efficiency—are typically only the byproduct of the (AI) techniques deployed at the implementation level, and are neither explicit object of study nor are taken into consideration at a requirements engineering phase. The paper presents a method, based on graph theory, to exactly compare and evaluate software design system configurations in the engineering of multiagent systems. The theoretical results are presented and validated on a crisis management scenario.

In-vehicle agents can potentially avert dangerous driving situations by adapting to the driver, context and traffic conditions. However, perceptions of system autonomy, the way agents offer assistance, driving contexts and users’ personality traits can all affect acceptance and trust. This paper reports on a survey-based experiment (N=100) that further investigates how these factors affect attitudes. The 2x2, between-subject, video-based design varied driving context (high, low density traffic) and type of agent (providing information, providing instructions). Both type of agent and traffic context affected attitudes towards the agent, with attitudes being most positive towards the instructive agent in a light traffic context. Participants scoring high on locus of control reported a higher intent to follow-up on the agent's instructions. Driving-related anxiety and aggression increased perceived urgency of the video scenario.

In this paper we present a decision making framework to enable agents to dynamically select a coordination mechanism. To demonstrate our approach we introduce an abstract task environment in which agents have to cooperate to achieve their goals. These agents are capable of using two coordination mechanisms, a centralized and a decentralized mechanism. We show how the decision making framework is operationalized in this abstract task environment. Furthermore, in an experiment we compare the performance of two static organizations with an organization in which agents have the ability to switch between coordination mechanisms. Results show that the ability to switch improves performance of the MAS.

Information Retrieval Systems and Information Filtering Systems need to cope with the problems posed by the widespread and pervasive diffusion of huge amounts of information of distinct nature and format (videos, images, maps, audio, texts), distinct genre and topics (geographic information, products' and organizations' descriptions, scientific research documentations, news, messages in social networks, etc.), and from etherogenous sources (universities, governamental institutions, private companies, and individuals). Moreover, the importance of taking into account the intentions and aims of both the information's authors and the users searching for information brings toward considering the context in which the search is carried out. The design of effective systems for information access should take into account the multi-dimensional nature of relevance; this can be done by exploiting several indicators and properties, such as topicality, location, trust, reputation, freshness, etc., all contributing to define the relevance of documents to personal information needs. This is a very hard and complex task due to several reasons that a large volume of research has addressed. Nowadays, the research concerned with defining effective systems for information access spans over several hot topics, such as document mining, user modelling, personalized topic detection and tracking, geographic information retrieval and sharing, Web discovery and mining, etc. Information access technologies, and in particular IR, are currently being used in many different application contexts that go far beyond the initial scope of their design. The special Track on Information Access and Retrieval (IAR) was first organized within the ACM International Symposium on Applied Computing in 2002, to the main aim of allowing researchers and practitioners in the IR field to present their experiences in the middle of the theory-practice spectrum of IR and IF, where many successful applications lie. This year the special track is in its fourteen edition in the context of SAC, and it includes 11 full papers (with an acceptance rate of 23%), and four poster papers. Each paper was peer reviewed by at least 3 members of the Program Committee. Our deep gratitude goes to the members of the Program Committee (a list of which can be found in these proceedings): their help has been invaluable for carrying out a high quality selection process. Four papers deal with the problem of groups identification, either to group users or web sites: the papers are entitled "GoSim : A tool for flexible Preference-Driven Group Formation" (by Yves Mouafo, Zoé Faget and Allel Hadjali), "Exploring graph topology via matrix factorization to improve Wikification " (by Raoni Ferreira, Marco Cristo and Maria da Graca Pimentel), "A Sentiment-Based Item Description Approach for kNN Collaborative Filtering " (by Rafael D'Addio and Marcelo Manzato), and "Classifying Websites by Industry Sector: A Study in Feature Design " (by Giacomo Berardi, Andrea Esuli, Tiziano Fagni and Fabrizio Sebastiani). Other four papers deal with problems related with optimizing Web searches, either in terms of efficiency or by personalization, or by context extraction. They are entitled "A Universal Topic Framework (UniZ) and Its Application in Online Search" (by Youngchul Cha, Keng-hao Chang, Hari Bommaganti, Ye Chen, Tak Yan, Bin Bi and Junghoo Cho), "Column-specific Context Extraction for Web Tables" (by Katrin Braunschweig, Maik Thiele, Julian Eberius and Wolfgang Lehner), "Towards a Feature-Rich Data Set for Personalized Access to Long-Tail Content " (by Christin Seifert, Jörg Schlötterer and Michael Granitzer), "BranchGuide: An Indexing Technique for Efficient, Lossless Processing of Branching Path Queries (by Talles Viana and Glêdson Elias) . The final three full papers deal with the problem of real time filtering and recommendation. The papers are titled "Leveraging temporal expressions to filter vital documents related to an entity (by Rafik Abbes, Karen Pinel-Sauvagnat, Nathalie Hernandez and Mohand Boughanem ), "On the Impact of Entity Linking in Microblog Real-Time Filtering" (by Giacomo Berardi, Diego Ceccarelli, Andrea Esuli and Diego Marcheggiani) and "Real-time Recommendations for User-Item Streams" (by Andreas Lommatzsch) .

A flexible model for evaluating soft query with unequal preferences in fuzzy databases is proposed. We assume that conditions with unequal preferences have an exclusive meaning like in the request "find a holiday accommodation such that big apartments are preferred to high rating hotels." In this case it is assumed that the aggregator of the soft conditions is an implicit OR. Conversely, conditions with unequal importance have an inclusive meaning, like in the query "find a house to rent that is cheap (most important), big (important), new (fairly important)." In this case the implicit aggregator is an AND. What we propose in this article is to model preferences as modifiers of the semantics of the evaluation function of the conditions. Because the soft conditions are aggregated by an OR, the more a soft condition is preferred, the more its evaluation function tolerates a greater undersatisfaction of the soft condition. The proposed approach is formalized by considering two alternative semantics of the evaluation function: the first semantics defines the evaluation function by means of a generalized fuzzy inclusion measure, and the second one as a generalized similarity measure. These functions are parameterized so that their modification is simply achieved by tuning the functions' parameters.

This paper presents a user survey-based analysis of the correlation between the users' willingness to personalize Web search and their social network usage patterns. The participants' responses to the survey questions enabled us to use a regression model for identifying the relationship between SNS variables and willingness to personalize Web search; the obtained results show that there is a strong relationship between willingness for personalized Web search and social network usage patterns. Finally, based on the findings of our survey we present some implications and directions for future work.

The aim of this contribution is to easily monitor the reputation of a company in the Twittersphere.We propose a strategy that organizes a stream of tweets into different clusters based on the tweets topics. Furthermore, the obtained clusters are prioritized into different priority levels. A cluster with high priority represents a topic which may affect the reputation of a company, and that consequently deserves immediate attention. The evaluation results show that our method is competitive even though the method does not make use of any external knowledge resource.

Social media repositories serve as a significant source of evidence when extracting information related to the reputation of a particular entity (e.g., a particular politician, singer or company). Reputation management experts manually mine the social media repositories (in particular Twitter) for monitoring the reputation of a particular entity. Recently, the online reputation management evaluation campaign known as RepLab at CLEF has turned attention to devising computational methods for facilitating reputation management experts. A quite significant research challenge related to the above issue is to disambiguate tweets with respect to entity names. In fact, finding if a particular tweet is relevant or irrelevant to a particular entity is an important task not satisfactorily solved yet; to address this issue in this paper we use \context phrases" in a tweet and Wikipedia disambiguated articles for a particular entity in an SVM classifier that utilizes features extracted from the Wikipedia graph structure i.e., links into Wikipedia articles and links from Wikipedia articles. Additionally we also use features derived from term-specificity and term-collocation features derived from the Wikipedia article of an entity under investigation. The experimental evaluations do not show a significant improvement over the baseline and other systems outperform our approach; however, manual inspection of feature sets and training data demonstrates the proposed Wikipedia graph-based features may show a promising outcome when used in combination with sophisticated learning algorithms

Extracting from a given document collection what we call "domain-specific" key terms/phrases is a challenging task. By "domain-specific" key terms/phrases we mean words/expressions representative of the topical areas specific to the focus of a document collection. For example, when a collection is related to academic research (i.e., its focus is related to topics dealing with academic research), the domain-specific key terms/phrases could be 'Information Retrieval', 'Marine Biology', 'Science', etc. In this contribution a technique for identifying domain-specific key terms/phrases from a collection of documents is proposed. The proposed technique works on short textual descriptions, and it makes use of the titles of Wikipedia articles and of the Wikipedia category graph. We performed some experiments over the document collection (html title text only) of eight post-graduate school Web sites of five different countries. The evaluations show promising results for the identification of domain-specific key terms/phrases.

The goal of this work is to formally provide a general definition of a multimedia recommendation system (MMRS), in particular a content-based MMRS (CB-MMRS), and to shed light on different applications of multimedia content for solving a variety of tasks related to recommendation. We would like to disambiguate the fact that multimedia recommendation is not only about recommending a particular media type (e.g., music, video), rather there exists a variety of other applications in which the analysis of multimedia input can be usefully exploited to provide recommendations of various kinds of information.

This paper presents a formal definition of an extension of the XQuery Full-Text language: The proposed extension consists in adding two new flexible axes, named below and near, which express structural constraints that can be specified by the user. Both constraints are evaluated in an approximate way with respect to a considered path, and their evaluation produces a path relevance score for each retrieved element. The formal syntax and the semantics of the two new axis are presented and discussed.

International audience; We present a new formal approach to retrieval personaliza- tion which emcompasses a query personalization process at the user's side with a light document adaptation at the in- formation server's side. Our solution relies on the use of a domain ontology: queries and documents are in fact indexed by sets of concepts. For each concept of the query, the query personalization process allows to express the importance of linked concepts, which may vary according to the search con- text. Each query concept can be ”clarified” by this process; although the proposed method clarifies only central query concepts. The initial query as well as its defined clarifica- tions are sent to the server. Then, the server reconsiders its document representations based on both the query and the concepts clarifications it received. The proposed solution does not require that the information server maintains any user profile, and can be useful when, for privacy concerns, it is committed not to profiling the users.