Search results

The purpose of this paper is to bring together the textual and multimedia opinions, since the use of social data has become the new trend that enables to gather the product reputation traded in social media. Integrating a product reputation process into the companies' strategy will bring several benefits such as helping in decision-making regarding the current and the new generation of the product by understanding the customers’ needs. However, image-centric sentiment analysis has received much less attention than text-based sentiment detection.

In this work, the authors propose a multimedia content-based product reputation framework that helps in detecting opinions from social media. Thus, in this case, the analysis of a certain publication is made by combining their textual and multimedia parts.

To test the effectiveness of the proposed framework, a case study based on YouTube videos has been established, as it brings together the image, the audio and the video processing at the same time.

The key novelty is the implication of multimedia content in addition of the textual one with the goal of gathering opinions about a certain product. The multimedia analysis brings together facial sentiment detection, printed text analysis, opinion detection from speeches and textual opinion analysis.

This paper aims to detect opinion leaders, who they play a vital role as influencers of their community, which will help companies to improve their image in social media. This idea came with the fast development of social media, where individuals are increasingly sharing their personal experiences, opinions and critiques about products through these platforms. Thus, the new customers can rely on these spontaneous recommendations to proceed with the purchase without risk of disappointment. Therefore, the mismanagement of the e-reputation can cause huge losses for companies.

In this study, a product reputation framework based on the prediction of opinion leaders is presented. To do so, opinion mining has been used to determine the product reputation in social media. In addition to posts processing, the profile information has also exploited to predict opinion leaders. To achieve the authors’ goal, spammers and duplicated profiles have been detected to improve the product reputation results.

The effectiveness of this approach has been tested using a social media simulation. The obtained results show that this approach is efficient and more accurate compared to the classical solutions.

The key novelty is the gathering of spammer detection criteria with different weights and the profiles matching by providing the suitable matching methods that take into account the profile’s attributes types. Consequently, a different similarity measure was assigned for each of the considered four attributes types. These two steps can ensure that the results obtained from social media are actually supported by opinions extracted directly from the real physical consumers.

In any critical system, high‐availability of software components like web services has so far been achieved through replication. Three replication strategies known as active, passive, and hybrid, describe for example how many replicas are needed, where to locate replicas, and how replicas interact with the original web service and among themselves if needed. The purpose of this paper is to show how replicates could be substituted with components that are similarly functional to the component that needs back‐up in case of failure.

After examination of the different existing replication strategies, it was decided to test the suitability of the proposed web services high‐availability approach based on communities for each strategy. To this end, the specification of web services using two behaviors, namely control and operational, was deemed appropriate.

The active replication strategy is the only strategy that could support the development of a web services high‐availability approach based on communities of web services.

The proposed approach has been validated in practice by deploying a JXTA‐based testbed. The experimental work has implemented the active replication strategy.

Software component high‐availability could be achieved by components that are similarly functional to this component, which permits the common limitations of existing replication strategies to be addressed.

The event‐driven paradigm is appropriate for context aware, distributed applications, yet basic events may be too low level to be meaningful to users. The authors aim to argue that this bottom‐up approach is insufficient to handle very low‐level sensor data or to express all the queries users might wish to make.

The authors propose an alternative model for querying and subscribing transparently to distributed state in a real‐time, ubiquitous, sensor‐driven environment such as is found in Sentient Computing.

The framework consists of four components: a state‐based, temporal first‐order logic (TFOL) model that represents the concrete state of the world, as perceived by sensors; an expressive TFOL‐based language, the Abstract Event Specification Language (AESL) for creating abstract event definitions, subscriptions and queries; a higherorder service (Abstract Event Detection Service) that accepts a subscription containing an abstract event definition as an argument and in return publishes an interface to a further service, an abstract event detector; and a satisfiability service that applies classical, logical satisfiability in order to check the satisfiability of the AESL definitions against the world model, in a manner similar to a constraint‐satisfaction problem.

The paper develops a model‐based approach, appropriate for distributed, heterogeneous environments.

People are subjected to a multitude of interruptions. In order to manage these interruptions it is imperative to predict a person's interruptability – his/her current readiness or inclination to be interrupted. This paper aims to introduce the approach of direct interruptability inference from sensor streams (accelerometer and audio data) in a ubiquitous computing setup and to show that it provides highly accurate and robust predictions.

The authors argue that scenarios are central for evaluating the performance of ubiquitous computing devices (and interruptability predicting devices in particular) and prove this on the setup employed, which was based on that of Kern and Schiele.

The paper demonstrates that scenarios provide the foundation for avoiding misleading results, and provide the basis for a stratified scenario‐based learning model, which greatly speeds up the training of such devices.

The direct prediction seems to be competitive or even superior to indirect prediction methods and no drawbacks have been observed yet.

The paper introduces a method for accurately predicting a person's interruptability directly from simple sensors without any intermediate steps/symbols.

The simple environment for context aware systems (SECAS) Project deals with the adaptation of applications to the context (user preferences and environment, terminal, etc.). The authors aim to develop a platform which makes the services, data and the user interface of applications adaptable to different context situations.

Previous research has concentrated on how to capture context data and how to carry it to the application. The present work focuses on the impact of context on the application core. A case study in the medical field is also analysed.

The paper illustrates a new definition of the context which separates the application data from the parameters of the context. This definition helps to establish a complete study on how to adapt applications on their three dimensions (services, content and presentation) to the context.

The paper presents the SECAS platform, one that ensures the deployment of adaptive context‐aware applications.

The behavior of tourists strongly depends on the availability and quality of information. Too little information as well as too much can be disorienting and forces many tourists to join the majority visiting major sights. This causes a few crowded places in contrast to many which are under‐utilized. A Destination Management Organization has the goal to spread tourists more evenly, whereas the tourists would like to enjoy the destination to its full potential according to their personal interests. The paper aims to focus on the issues surrounding the creation of a mobile tourist guide.

A field trial was conducted in the summer of 2005 to study the following questions as a precondition for the development: Is it possible to seed generic interest profiles in the mobile context that allow the accurate prediction of actual rankings? Are the interest profiles sufficiently diverse to base personalized tours on individual interest profiles instead of interest prototypes? How do personalized tours affect the spatial behavior of tourist? Three methods to elicit the generic preferences of tourist in the mobile context are compared with actual rankings using Spearman's rank order coefficient.

The diversity of the interest profiles is analyzed in various ways leading to the conclusion that personalized interest profiles are necessary. For the gathered profiles tours are computed and simulated in order to gain a first insight into the effect on the spatial behavior of tourists.

The dynamic tour guide is supporting both goals by means of pervasive computing based on the actual context which is defined by personal interests, location and schedule of a tourist. It enables a personalized, spontaneous and guided tour.

A context‐aware system is a system that uses context to provide relevant information or services to its users. While there has been a variety of context middleware infrastructures and context‐aware applications, little work has been done on integrating context into database management systems. The purpose of this paper is to consider a preference database system that supports context‐aware queries, that is, queries whose results depend on the context at the time of their submission.

The paper proposes using data cubes to store the dependencies between context‐dependent preferences and database relations and on‐line analytical processing techniques for processing context‐aware queries. This allows for the manipulation of the captured context data at various levels of abstraction, for instance, in the case of a context parameter representing location, preferences can be expressed, for example, at the level of a city, the level of a country or both. To improve query performance, the paper uses an auxiliary data structure, called context tree. The context tree stores results of past context‐aware queries indexed by the context of their execution. Finally, the paper outline the implementation of a prototype context‐aware restaurant recommender.

The use of context is important in many applications such as pervasive computing where it is important that users receive only relevant information.

Although there is much research on location‐aware query processing in the area of spatial‐temporal databases, integrating other forms of context in query processing is a rather new research topic.