1,720,972 research outputs found
A principled approach to context schema evolution in a data management perspective
Full text linkContext-aware data tailoring studies the means for the system to furnish the users, at any moment, only with the set of data which is relevant for their current context. These data may be from traditional databases, sensor readings, environmental information, close-by people, points of interest, etc. To implement context-awareness, we use a formal representation of a conceptual context model, used to design the context schema, which intensionally represents all the contexts in which the user may be involved in the considered application scenario.
Following this line of thought, in this paper we develop a formal approach and the corresponding strategy to manage the evolution of the context schema of a given context-aware application, when the context perspectives initially envisaged by the system designer are not applicable any more and unexpected contexts are to be activated. Accordingly, when the context schema evolves also the evolution of the corresponding context-aware data portions must be taken care of. The aim of this paper is thus to provide the necessary conceptual and formal notions to manage the evolution of a context schema in the perspective of data tailoring: after introducing a set of operators to manage evolution and proving their soundness and completeness, we analyze the impact that context evolution has on the context-based data tailoring process. We then study how sequences of operator applications can be optimized and finally present a prototype validating the feasibility of the approach
Long-term correlations in short, non-stationary time series: An application to international R&D collaborations
Full text linkWithin the perimeter of patent collaboration networks, the average distance of collaborations and the number of countries involved per each collaboration have been shown to have increased steadily in time. Less attention, though, has been devoted to assessing whether this growth of cross-country collaborations is stable in time. To address this scientific question we focus on the identification of long-term correlations (i.e. temporal persistence). Our data set consists of time series of yearly average collaboration radii and of cross-border links in the Euro-American subsystem of the global collaboration network for the period 1978-2014. To investigate the fundamental persistence properties of these time series, we use Detrended Fluctuation Analysis, a method that allows us to measure long-term correlations in detrended signals. Also, we devise a general and original procedure to assess the statistical significance of results for short time series. Our results, showing that long-term correlations do exist in the majority of our signals, reinforce the hypothesis of a diminishing role of geographical distance in technological collaborations. Results at national level show that a significant degree of heterogeneity in persistence parameters can be detected within Europe, irrespectively of the efforts towards the set-up of an integrated European Research Area
Recommending New Items to Ephemeral Groups Using Contextual User Influence
No full textGroup recommender systems help groups of users in finding appropriate items to be enjoyed together. Lots of activities, like watching TV or going to the restaurant, are intrinsically
group-based, thus making the group recommendation problem very relevant. In this paper we study ephemeral groups, i.e., groups where the members might be together for the first time. Recent approaches have tackled this issue introducing complex models to be learned offline, making them unable to deal with new items; on the contrary, we propose a group recommender able to manage new items too. In more detail, our technique determines the preference of a group for an item by combining the individual preferences of the group members on the basis of their contextual influence, where the contextual influence represents the ability of an individual, in a given situation, to direct the group’s decision. We conducted an extensive experimental evaluation on a TV dataset containing a log of viewings performed by real groups, showing how our approach outperforms the comparable techniques from the literature
Mining Context-Aware Preferences on Relational and Sensor Data
No full textThe increasing amount of available digital data motivates the development of techniques for the management of the information overload which risks to actually reduce people’s knowledge instead of increasing it. Research is concentrating on topics related to the problem of filtering/suggesting a subset of available information that is likely to be of interest to the user, besides this subset may vary and is often determined by the context the user is currently in. We cannot actually expect only a collaborative approach, where users manually specify the long list of preferences that might be applied to all available data; that is why in this paper we propose a preliminary methodology, described by using a realistic running example, that tries to combine the following research topics: context-awareness, data mining, and preferences. In particular, data mining is used to infer contextual preferences from the previous user’s querying activity on static data and on available dynamic values coming from sensors
Discovering Contextual Association Rules in Relational Databases
No full textContextual association rules represent co-occurrences between contexts and properties of data, where the context is a set of environmental or user personal features employed to customize an application. Due to their particular structure, these rules can be very tricky to mine, and if the process is not carried out with care, an unmanageable set of not significant rules may be extracted. In this paper we survey two existing algorithms for relational databases and present a novel algorithm that merges the two proposals overcoming their limitations
Disjunctive constraints in RDF and their application to context schemas
No full textRDF is a data model whose relevance is growing in the last years. Recently, some proposals have enriched the model with integrity constraints, well known within relational databases. In this work we extend an existing framework with two new types of integrity constraints of disjunctive nature, inspired by similar kinds of dependencies studied for the relational model. The problem of the logical implication for the two novel categories is also analyzed. Moreover, as an application scenario, we propose a complete and independent set of constraints to model the context in RDF, where the context is a notion employed in databases to perform information filtering on the basis of the user's current situation
Efficiently using contextual influence to recommend new items to ephemeral groups
Full text linkGroup recommender systems suggest items to groups of users that want to utilize those items together. These systems can support several activities that can be performed together with other people and are typically social, like watching TV or going to the restaurant. In this paper we study ephemeral groups, i.e., groups constituted by users who are together for the first time, and for which therefore there is no history of past group activities.Recent works have studied ephemeral group recommendations proposing techniques that learn complex models of users and items. These techniques, however, are not appropriate to recommend items that are new in the system, while we propose a method able to deal with new items too. Specifically, our technique determines the preference of a group for a given item by combining the individual preferences of the group members on the basis of their contextual influence, the contextual influence representing the ability of an individual, in a given situation, to guide the group's decision. Moreover, while many works on recommendations do not consider the problem of efficiently producing recommendation lists at runtime, in this paper we speed up the recommendation process by applying techniques conceived for the top-K query processing problem. Finally, we present extensive experiments, evaluating: (i) the accuracy of the recommendations, using a real TV dataset containing a log of viewings performed by real groups, and (ii) the efficiency of the online recommendation task, exploiting also a bigger partially synthetic dataset. (C) 2019 Elsevier Ltd. All rights reserved
Context Schema Evolution in Context-Aware Data Management
No full textPervasive access - often by means of mobile devices - to the massive amount of available (Web) data suggests to deliver, anywhere at any time, exactly the data that are needed in the current specific situation. The literature has introduced the possibility to describe the context in which the user is involved, and to tailor the available data on its basis. In this paper, after having formally defined the context schema - a representation for the contexts which are to be expected in a given
application scenario - a strategy to manage context schema evolution is developed, by introducing a sound and complete set of operators
Reducing Big Data by Means of Context-Aware Tailoring
No full textContext-aware personalization is one of the possible ways to face the problem of information overload, that is, the difficulty of understanding an issue and making decisions when receiving too much information. Context-aware personalization can reduce the information noise, by proposing to the users only the information which is relevant to their current contexts. In this work we propose an approach that uses data mining algorithms to automatically infer the subset of data that, for each context, must be presented to the user, thus reducing the information noise
- …
