Search

In this article, we propose a system able to implicitly assess a user’s expertise in a particular topic based on her publications (e.g., scientific papers) on it and available through online bibliographic databases. This task is performed through two different approaches, both of them based on a graph-based model. The first approach (content-based) considers the text content, the second one (collaborative) analyzes the relationships in the same content in terms of co-citations. Preliminary experimental results are encouraging and raise several interesting considerations. In particular, they show that the best solution is obtained by integrating the two approaches above, in which each of them allows the system to overcome the limitations of the other one.

Nowadays, several location-based services (LBSs) allow their users to take advantage of information from the Web about points of interest (POIs) such as cultural events or restaurants. To the best of our knowledge, however, none of these provides information taking into account user preferences, or other elements, in addition to location, that contribute to define the context of use. The provided suggestions do not consider, for example, time, day of week, weather, user activity or means of transport. This article describes a social recommender system able to identify user preferences and information needs, thus suggesting personalized recommendations related to POIs in the surroundings of the user's current location. The proposed approach achieves the following goals: (i) to supply, unlike the current LBSs, a methodology for identifying user preferences and needs to be used in the information filtering process; (ii) to exploit the ever-growing amount of information from social networking, user reviews, and local search Web sites; (iii) to establish procedures for defining the context of use to be employed in the recommendation of POIs with low effort. The flexibility of the architecture is such that our approach can be easily extended to any category of POI. Experimental tests carried out on real users enabled us to quantify the benefits of the proposed approach in terms of performance improvement.

Recommender systems help online users find relevant content by suggesting information of potential interest to them [4]. Social recommender is any recommender with online social relations as an additional input, namely, augmenting an existing recommendation engine with additional social content [5]. In this talk we describe our experience and lessons learned in developing social recommender systems able to deliver attractive and relevant content. More specifically, we focus on machine learning and data mining techniques exploited for the following goals: (i) to extract user preferences and needs to be used in the information filtering process; (ii) to harness the vast amount of information from user reviews, social networking, and local search Web sites; (iii) to infer peculiar users’ attitudes (i.e., sentiments, opinions, and ways of thinking) toward their own interests [3]; (iv) to define the context of use in the recommendation process [1]. Achieving the above goals allowed us to realize a social recommender system for context-aware mobile services [2], which provide users with personalized recommendations about points of interest (e.g., restaurants or cultural events) in the surroundings of the user’s current position.

Classic query expansion approaches are based on the use of two-dimensional co-occurrence matrices. In this paper, we propose the adoption of three-dimensional matrices, where the added dimension is represented by semantic classes (i.e., categories comprising all the terms that share a semantic property) related to the folksonomy extracted from social bookmarking services, such as Delicious and StumbleUpon. The results of an in-depth experimental evaluation performed on real users show that our approach outperforms traditional techniques, so confirming the validity and usefulness of the categorization of the user needs and preferences in semantic classes.

In the last decade, social bookmarking services have gained popularity as a way of annotating and categoriz- ing a variety of different web resources. The idea behind this work is to exploit such services for enhancing traditional query expansion techniques. Specifically, the system we propose relies on three-dimensional co- occurrence matrices, where the further dimension is introduced to represent categories of terms sharing the same semantic property. Such categories, named semantic classes, are related to the folksonomy mined from social bookmarking services such as Delicious, Digg, and StumbleUpon. The paper illustrates a comparative experimental evaluation on real datasets, such as the one collected by the Open Directory Project and the TREC 2004. We also include the results of a specific disambiguation analysis aimed to evaluate the effective- ness of our approach in comparison with state-of-the-art techniques when satisfying queries characterized by polysemic and ambiguous terms.

Weak semantic techniques rely on the integration of Semantic Web techniques with social annotations and aim to embrace the strengths of both. In this article, we propose a novel weak semantic technique for query expansion. Traditional query expansion techniques are based on the computation of two-dimensional co-occurrence matrices. Our approach proposes the use of three-dimensional matrices, where the added dimension is represented by semantic classes (i.e., categories comprising all the terms that share a semantic property) related to the folksonomy extracted from social bookmarking services, such as delicious and StumbleUpon . The results of an indepth experimental evaluation performed on both artificial datasets and real users show that our approach outperforms traditional techniques, such as relevance feedback and personalized PageRank, so confirming the validity and usefulness of the categorization of the user needs and preferences in semantic classes. We also present the results of a questionnaire aimed to know the users opinion regarding the system. As one drawback of several query expansion techniques is their high computational costs, we also provide a complexity analysis of our system, in order to show its capability of operating in real time.

Adaptive query expansion (QE) allows users to better define their search domain by supplementing the original query with additional terms related to their preferences and information needs. The system we present is an extension of the traditional QE techniques, which rely on the computation of two-dimensional co-occurrence matrices. Our system makes use of three-dimensional co-occurrence matrices, where the added dimension is represented by semantic classes (i.e., categories comprising all the terms that share a semantic property) related to the folksonomy extracted from social book marking services such as delicious, Digg, and Stumble Upon. The results of an indepth experimental evaluation on artificial datasets and real users show that our system outperforms some well-known approaches in the literature, as well as a state-of-the-art search engine.

Adaptive query expansion (QE) allows users to better define their search domain by supplementing the original query with additional terms related to their preferences and information needs. The system we present is an extension of the traditional QE techniques, which rely on the computation of two-dimensional co-occurrence matrices. Our system makes use of three-dimensional co-occurrence matrices, where the added dimension is represented by semantic classes (i.e., categories comprising all the terms that share a semantic property) related to the folksonomy extracted from social book-marking services such as delicious, Digg, and StumbleUpon. The generation of the user profile occurs through the creation of a model that is dynamically updated using the information gleaned from the searches (visited pages and corresponding search queries). The system analyzes the input queries and, if they actually reflect the interests already shown by the user in previous searches, it returns different QEs involving different semantic fields. The output of the system is structured in different blocks categorized through keywords, thus helping the user judge which result is most relevant to him. The results of an experimental evaluation involving real users are reported.

Traditional desktop search paradigm often does not fit mobile contexts. Common mobile devices provide impoverished mechanisms for text entry and small screens are able to offer only a limited set of options, therefore the users are not usually able to specify their needs. On a different note, mobile technologies have become part of the everyday life as shown by the estimate of one billion of mobile broadband subscriptions in 2011. This paper describes an approach to make context-aware mobile interaction available in scenarios where users might be looking for categories of points of interest (POIs), such as cultural events and restaurants, through remote location-based services. Empirical evaluations shows how rich representations of user contexts has the chance to increase the relevance of the retrieved POIs.

Most of the existing recommendation engines do not take into consideration contextual information for suggesting interesting items to users. Features such as time, location, or weather, may affect the user preferences for a particular item. In this paper, we propose two different context-aware approaches for the movie recommendation task. The first is an hybrid recommender that assesses available contextual factors related to time in order to increase the performance of traditional CF approaches. The second approach aims at identifying users in a household that submitted a given rating. This latter approach is based on machine learning techniques, namely, neural networks and majority voting classifiers. The effectiveness of both the approaches has been experimentally validated using several evaluation metrics and a large dataset.

Each business company collects, produces and exploits for its activities and goals large amounts of information. Most of the times this knowledge makes the intellectual capital for creating value and innovation. Knowledge management (KM) systems aim at manipulating knowledge by storing and redistributing corporate information that are acquired from the organization's members. In this context, Virtual Enterprises (VE) plays a crucial role as not permanent alliances of enterprises joined together to share resources and skills in order to better respond to business opportunities. The representation and retrieval of distributed knowledge is an important feature that information systems must provide in order to obtain advantages from this kind of enterprises. KEEN 1 (Knowledge-based Extended Enterprise) is a research project for developing a system able to extract and let different business companies access to collective knowledge required to achieve particular shared goals. In this paper, we report the most important features of this system, especially in the context of distributed knowledge representation and retrieval.

The Web has grown from a simple hypertext system for research labs to an ubiquitous information system including virtually all human knowledge, e.g., movies, images, music, documents, etc. The traditional browsing activity seems to be often inadequate to locate information satisfying the user needs. Even search engines, based on the Information Retrieval approach, with their huge indexes show many drawbacks, which force users to sift through long lists of results or reformulate queries several times. Recently, an important research activity effort has been focusing on this vast amount of machine-accessible knowledge and on how it can be exploited in order to match the user needs. The personalization and adaptation of the human-computer interaction in information seeking by means of machine learning techniques and in AI-based representations of the information help users to address the overload problem. This chapter illustrates the most important approaches proposed to personalize the access to information, in terms of gathering resources related to given topics of interest and ranking them as a function of the current user needs and activities, as well as examples of prototypes and Web systems.

Recommender Systems provide suggestions for items (e.g., movies or songs) to be of use to a user. They must take into account information to deliver more useful (perceived) recommendations. Current music recommender takes an initial input of a song and plays music with similar characteristics, or music that other users have listened to along with the input song. Listening behaviors in terms of temporal information associated to ratings or playbacks are usually ignored. We propose a recommender that predicts the most rated songs that a given user is likely to play in the future analyzing and comparing user listening habits by means of signal processing techniques. Recommender systems provide suggestions based on user preferences in order to recommend items likely to be of interest to a user. It is obvious that user preferences are influenced by the current context, such as the current time of the day, mood, or current activities. Nevertheless, a few recommender systems explicitly include this information in the preference models. A special group of recommender systems are the ones based on the collaborative approach (Resnick et al., 1994; Shardanand and Maes, 1995; Breese et al., 1998). The system generates recommendations using only information about rating profiles for different users. Collaborative systems locate peer users with a rating history similar to the current user and generate recommendations using this neighborhood. Collaborative filtering (CF) systems have been successful in several recommender systems. The availability of large datasets and additional information that is easy collectable from the web, makes this task interesting. There are several issues that do not allow us to directly apply the traditional CF approach for music recommendation. The space of possible items (i.e., tracks) can be very large and, similarly, the user space can also be enormous. Often user ratings are not available or they cover only a small subset of the user library of songs. Moreover, when new users enter to the system or new songs are added to the global library, it is not possible to provide any recommendation to them due to the lack of any preference information (the so known cold-start problem). There is no chance to use taxonomies or ontologies to represent the new items and facilitate the clustering as happens in different domains (e.g., (Acampora et al., 2010a; Micarelli et al., 2009)) Content-based approaches collect information describing the items and then, based on the user preferences, they predict which tracks the user might enjoy (see for example the Pandora service1). The key component of this approach is the similarity function among the songs. Nevertheless, there is a strong limitation of the highlevel descriptors that can be automatically extracted from the tracks (Celma, 2010). One more relevant issue that traditional CF approaches do not take into consideration is the listening behavior of the user in terms of temporal information. The timestamp of an item (i.e., when the song song is played) is an important factor for the recommendation algorithm. Usually, the prediction function treats the older items as less relevant than the new ones, but any further reasoning about the temporal information is simply ignored. In this paper, we discuss a recommendation approach based on signal processing. In particular, a traditional CF approach is enhanced considering an improved similarity function between users. The user listening habits are represented by signals. Wavelet theory is used to study the related time-frequency representations of signals and draw similarity between listening behaviors. Signal processing techniques are not employed to extract features from the songs, but for representing and comparing those behaviors in or

Each business company collects, produces and exploits for its activities and goals large amounts of information. Most of the times this knowledge makes the intellectual capital for creating value and innovation. Knowledge management (KM) systems aim at manipulating knowledge by storing and redistributing corporate information that are acquired from the organizations members. In this context, Virtual Enterprises (VE) plays a crucial role as not permanent alliances of enterprises joined together to share resources and skills in order to better respond to business opportunities. The representation and retrieval of distributed knowledge is an important feature that information systems must provide in order to obtain advantages from this kind of enterprises. PVE (Personalized Virtual Enterprise) is an ongoing research project for developing a system able to extract and let different business companies access to collective knowledge required to achieve particular shared goals. In this paper, we report the most important features of this system, especially in the context of distributed knowledge representation and retrieval.