Bunker-de-bunk 2012

Tourist snapshot : the mediating gaze of the Sydney Opera House

There has already been much scholarly work produced about the Sydney Opera House, discussing the production of its architecture, historical and political context, and its symbolic meaning to Sydney. However little scholarly attention has been paid to the way this building is represented through tourist practices of photography. The essay attempts to bring an architectural perspective to the study of this tourist practice, which is usually addressed from the disciplines of cultural geography, sociology and anthropology. The essay considers the above questions by the analysis of some 300 images sourced from the photo sharing website Flickr (www.flickr.com). It draws on John Urry's notion of the 'tourist gaze' which describes how places are structured and regulated by the visual. The essay then uses Jonas Larsen's work, which position tourist photography as a performance of social relations to argue that the activity of photographing the Sydney Opera House is more than a ritual of consumption, and can be seen as an embodied performance located at the intersection of space, experience and image.

A differentially private algorithm for location data release

The rise of mobile technologies in recent years has led to large volumes of location information, which are valuable resources for knowledge discovery such as travel patterns mining and traffic analysis. However, location dataset has been confronted with serious privacy concerns because adversaries may re-identify a user and his/her sensitivity information from these datasets with only a little background knowledge. Recently, several privacy-preserving techniques have been proposed to address the problem, but most of them lack a strict privacy notion and can hardly resist the number of possible attacks. This paper proposes a private release algorithm to randomize location dataset in a strict privacy notion, differential privacy, with the goal of preserving users’ identities and sensitive information. The algorithm aims to mask the exact locations of each user as well as the frequency that the user visits the locations with a given privacy budget. It includes three privacy-preserving operations: private location clustering shrinks the randomized domain and cluster weight perturbation hides the weights of locations, while private location selection hides the exact locations of a user. Theoretical analysis on privacy and utility confirms an improved trade-off between privacy and utility of released location data. Extensive experiments have been carried out on four real-world datasets, GeoLife, Flickr, Div400 and Instagram. The experimental results further suggest that this private release algorithm can successfully retain the utility of the datasets while preserving users’ privacy.

Informação na Web colaborativa: um olhar para o direito autoral e as alternativas emergentes

Pós-graduação em Ciência da Informação - FFC

Strategie di elaborazione di tracciati GPS per reti veicolari

Il proliferare di dispositivi di elaborazione e comunicazione mobili (telefoni cellulari, computer portatili, PDA, wearable devices, personal digital assistant) sta guidando un cambiamento rivoluzionario nella nostra società dell'informazione. Si sta migrando dall'era dei Personal Computer all'era dell'Ubiquitous Computing, in cui un utente utilizza, parallelamente, svariati dispositivi elettronici attraverso cui può accedere a tutte le informazioni, ovunque e quantunque queste gli si rivelino necessarie. In questo scenario, anche le mappe digitali stanno diventando sempre più parte delle nostre attività quotidiane; esse trasmettono informazioni vitali per una pletora di applicazioni che acquistano maggior valore grazie alla localizzazione, come Yelp, Flickr, Facebook, Google Maps o semplicemente le ricerche web geo-localizzate. Gli utenti di PDA e Smartphone dipendono sempre più dai GPS e dai Location Based Services (LBS) per la navigazione, sia automobilistica che a piedi. Gli stessi servizi di mappe stanno inoltre evolvendo la loro natura da uni-direzionale a bi-direzionale; la topologia stradale è arricchita da informazioni dinamiche, come traffico in tempo reale e contenuti creati dagli utenti. Le mappe digitali aggiornabili dinamicamente sono sul punto di diventare un saldo trampolino di lancio per i sistemi mobili ad alta dinamicità ed interattività, che poggiando su poche informazioni fornite dagli utenti, porteranno una moltitudine di applicazioni innovative ad un'enorme base di consumatori. I futuri sistemi di navigazione per esempio, potranno utilizzare informazioni estese su semafori, presenza di stop ed informazioni sul traffico per effettuare una ottimizzazione del percorso che valuti simultaneamente fattori come l'impronta al carbonio rilasciata, il tempo di viaggio effettivamente necessario e l'impatto della scelta sul traffico locale. In questo progetto si mostra come i dati GPS raccolti da dispositivi fissi e mobili possano essere usati per estendere le mappe digitali con la locazione dei segnali di stop, dei semafori e delle relative temporizzazioni. Queste informazioni sono infatti oggi rare e locali ad ogni singola municipalità, il che ne rende praticamente impossibile il pieno reperimento. Si presenta quindi un algoritmo che estrae utili informazioni topologiche da agglomerati di tracciati gps, mostrando inoltre che anche un esiguo numero di veicoli equipaggiati con la strumentazione necessaria sono sufficienti per abilitare l'estensione delle mappe digitali con nuovi attributi. Infine, si mostrerà come l'algoritmo sia in grado di lavorare anche con dati mancanti, ottenendo ottimi risultati e mostrandosi flessibile ed adatto all'integrazione in sistemi reali.

Creative Commons Olympics. How Big Media is Learning to License From Amateur Authors

NBC Universal’s decision to use Creative Commons-licensed photographs in an Olympic broadcast is an example of how media conglomerates are experimenting with collaboration with amateurs, but it also reveals potential problems of letting non-lawyers negotiate copyright licensing agreements. In the process, NBC’s producers nearly opened the door for a multimillion-dollar infringement law suit. To avoid such pitfalls, media companies need to adopt policies and best practices for using amateur licensed works. These guidelines should instruct how a production can attribute collaborating authors and how the Open Content licensing terms affect the licensing of the productions. The guidelines should also instruct how producers can seek alternative licensing arrangements with amateurs and contribute back to the Open Content community.

Customized mashups for improved reputation visualization

This document describes a possible use for the YouReputation API. A mashup combining the YouReputation and the Flickr APIs attempts to improve the visualization of reputation. First, this paper gives an introduction to Web services and APIs and further explains the developed prototype. This paper introduces an API that can be easily combined with other APIs to improve the representation of reputation terms and therefore enhance usability and design.

Multipedia: Enriching DBpedia with Multimedia Information

Enriching knowledge bases with multimedia information makes it possible to complement textual descriptions with visual and audio information. Such complementary information can help users to understand the meaning of assertions, and in general improve the user experience with the knowledge base. In this paper we address the problem of how to enrich ontology instances with candidate images retrieved from existing Web search engines. DBpedia has evolved into a major hub in the Linked Data cloud, interconnecting millions of entities organized under a consistent ontology. Our approach taps into the Wikipedia corpus to gather context information for DBpedia instances and takes advantage of image tagging information when this is available to calculate semantic relatedness between instances and candidate images. We performed experiments with focus on the particularly challenging problem of highly ambiguous names. Both methods presented in this work outperformed the baseline. Our best method leveraged context words from Wikipedia, tags from Flickr and type information from DBpedia to achieve an average precision of 80%.

A framework for building distributed social network websites

Las redes son la esencia de comunidades y sociedades humanas; constituyen el entramado en el que nos relacionamos y determinan cómo lo hacemos, cómo se disemina la información o incluso cómo las cosas se llevan a cabo. Pero el protagonismo de las redes va más allá del que adquiere en las redes sociales. Se encuentran en el seno de múltiples estructuras que conocemos, desde las interaciones entre las proteínas dentro de una célula hasta la interconexión de los routers de internet. Las redes sociales están presentes en internet desde sus principios, en el correo electrónico por tomar un ejemplo. Dentro de cada cliente de correo se manejan listas contactos que agregadas constituyen una red social. Sin embargo, ha sido con la aparición de los sitios web de redes sociales cuando este tipo de aplicaciones web han llegado a la conciencia general. Las redes sociales se han situado entre los sitios más populares y con más tráfico de la web. Páginas como Facebook o Twitter manejan cifras asombrosas en cuanto a número de usuarios activos, de tráfico o de tiempo invertido en el sitio. Pero las funcionalidades de red social no están restringidas a las redes sociales orientadas a contactos, aquellas enfocadas a construir tu lista de contactos e interactuar con ellos. Existen otros ejemplos de sitios que aprovechan las redes sociales para aumentar la actividad de los usuarios y su involucración alrededor de algún tipo de contenido. Estos ejemplos van desde una de las redes sociales más antiguas, Flickr, orientada al intercambio de fotografías, hasta Github, la red social de código libre más popular hoy en día. No es una casualidad que la popularidad de estos sitios web venga de la mano de sus funcionalidades de red social. El escenario es más rico aún, ya que los sitios de redes sociales interaccionan entre ellos, compartiendo y exportando listas de contactos, servicios de autenticación y proporcionando un valioso canal para publicitar la actividad de los usuarios en otros sitios web. Esta funcionalidad es reciente y aún les queda un paso hasta que las redes sociales superen su condición de bunkers y lleguen a un estado de verdadera interoperabilidad entre ellas, tal como funcionan hoy en día el correo electrónico o la mensajería instantánea. Este trabajo muestra una tecnología que permite construir sitios web con características de red social distribuída. En primer lugar, se presenta una tecnología para la construcción de un componente intermedio que permite proporcionar cualquier característica de gestión de contenidos al popular marco de desarrollo web modelo-vista-controlador (MVC) Ruby on Rails. Esta técnica constituye una herramienta para desarrolladores que les permita abstraerse de las complejidades de la gestión de contenidos y enfocarse en las particularidades de los propios contenidos. Esta técnica se usará también para proporcionar las características de red social. Se describe una nueva métrica de reusabilidad de código para demostrar la validez del componente intermedio en marcos MVC. En segundo lugar, se analizan las características de los sitios web de redes sociales más populares, con el objetivo de encontrar los patrones comunes que aparecen en ellos. Este análisis servirá como base para definir los requisitos que debe cumplir un marco para construir redes sociales. A continuación se propone una arquitectura de referencia que proporcione este tipo de características. Dicha arquitectura ha sido implementada en un componente, Social Stream, y probada en varias redes sociales, tanto orientadas a contactos como a contenido, en el contexto de una asociación vecinal tanto como en proyectos de investigación financiados por la UE. Ha sido la base de varios proyectos fin de carrera. Además, ha sido publicado como código libre, obteniendo una comunidad creciente y está siendo usado más allá del ámbito de este trabajo. Dicha arquitectura ha permitido la definición de un nuevo modelo de control de acceso social que supera varias limitaciones presentes en los modelos de control de acceso para redes sociales. Más aún, se han analizado casos de estudio de sitios de red social distribuídos, reuniendo un conjunto de caraterísticas que debe cumplir un marco para construir redes sociales distribuídas. Por último, se ha extendido la arquitectura del marco para dar cabida a las características de redes sociales distribuídas. Su implementación ha sido validada en proyectos de investigación financiados por la UE. Abstract Networks are the substance of human communities and societies; they constitute the structural framework on which we relate to each other and determine the way we do it, the way information is diseminated or even the way people get things done. But network prominence goes beyond the importance it acquires in social networks. Networks are found within numerous known structures, from protein interactions inside a cell to router connections on the internet. Social networks are present on the internet since its beginnings, in emails for example. Inside every email client, there are contact lists that added together constitute a social network. However, it has been with the emergence of social network sites (SNS) when these kinds of web applications have reached general awareness. SNS are now among the most popular sites in the web and with the higher traffic. Sites such as Facebook and Twitter hold astonishing figures of active users, traffic and time invested into the sites. Nevertheless, SNS functionalities are not restricted to contact-oriented social networks, those that are focused on building your own list of contacts and interacting with them. There are other examples of sites that leverage social networking to foster user activity and engagement around other types of content. Examples go from early SNS such as Flickr, the photography related networking site, to Github, the most popular social network repository nowadays. It is not an accident that the popularity of these websites comes hand-in-hand with their social network capabilities The scenario is even richer, due to the fact that SNS interact with each other, sharing and exporting contact lists and authentication as well as providing a valuable channel to publize user activity in other sites. These interactions are very recent and they are still finding their way to the point where SNS overcome their condition of data silos to a stage of full interoperability between sites, in the same way email and instant messaging networks work today. This work introduces a technology that allows to rapidly build any kind of distributed social network website. It first introduces a new technique to create middleware that can provide any kind of content management feature to a popular model-view-controller (MVC) web development framework, Ruby on Rails. It provides developers with tools that allow them to abstract from the complexities related with content management and focus on the development of specific content. This same technique is also used to provide the framework with social network features. Additionally, it describes a new metric of code reuse to assert the validity of the kind of middleware that is emerging in MVC frameworks. Secondly, the characteristics of top popular SNS are analysed in order to find the common patterns shown in them. This analysis is the ground for defining the requirements of a framework for building social network websites. Next, a reference architecture for supporting the features found in the analysis is proposed. This architecture has been implemented in a software component, called Social Stream, and tested in several social networks, both contact- and content-oriented, in local neighbourhood associations and EU-founded research projects. It has also been the ground for several Master’s theses. It has been released as a free and open source software that has obtained a growing community and that is now being used beyond the scope of this work. The social architecture has enabled the definition of a new social-based access control model that overcomes some of the limitations currenly present in access control models for social networks. Furthermore, paradigms and case studies in distributed SNS have been analysed, gathering a set of features for distributed social networking. Finally the architecture of the framework has been extended to support distributed SNS capabilities. Its implementation has also been validated in EU-founded research projects.

Sistema de pruebas de servicios REST mediante análisis de esquemas inferidos

A lo largo de los últimos años, el paradigma de la arquitectura orientada a servicios ha tenido una gran expansión gracias a la expansión de las tecnologías web e internet. Las ventajas de esta arquitectura se basan en ofrecer diseños modulares con poco acoplamiento entre sí, lo que permite la creación eficiente y sistemática de sistemas distribuidos. Para que este tipo de arquitectura sea posible, es necesario dotar a los servicios de interfaces de interconexión que permitan encapsular los servicios al mismo tiempo que faciliten el uso de los mismos. Existen varias tecnologías para definir estos interfaces. Entre ellas, los servicios REST, o REpresentional State Transfer, están logrando cada vez más aceptación. Esto se debe principalmente a su capacidad de escalabilidad y la uniformidad de sus interfaces, que permite una mayor separación entre los consumidores y los servicios. De hecho, compañias como Yahoo, Google o Twitter definen interfaces REST de acceso a sus servicios, ya se para consultar mapas (GoogleMaps), imágenes (Flickr) o el correo, permitiendo a terceros desarrollar clientes para sus servicios sin tener que involucrarse en su producción.

Exploring the World's Visual History

Thesis (Ph.D.)--University of Washington, 2016-08

Towards Proactive Context-aware Computing and Systems

A primary goal of context-aware systems is delivering the right information at the right place and right time to users in order to enable them to make effective decisions and improve their quality of life. There are three key requirements for achieving this goal: determining what information is relevant, personalizing it based on the users’ context (location, preferences, behavioral history etc.), and delivering it to them in a timely manner without an explicit request from them. These requirements create a paradigm that we term as “Proactive Context-aware Computing”. Most of the existing context-aware systems fulfill only a subset of these requirements. Many of these systems focus only on personalization of the requested information based on users’ current context. Moreover, they are often designed for specific domains. In addition, most of the existing systems are reactive - the users request for some information and the system delivers it to them. These systems are not proactive i.e. they cannot anticipate users’ intent and behavior and act proactively without an explicit request from them. In order to overcome these limitations, we need to conduct a deeper analysis and enhance our understanding of context-aware systems that are generic, universal, proactive and applicable to a wide variety of domains. To support this dissertation, we explore several directions. Clearly the most significant sources of information about users today are smartphones. A large amount of users’ context can be acquired through them and they can be used as an effective means to deliver information to users. In addition, social media such as Facebook, Flickr and Foursquare provide a rich and powerful platform to mine users’ interests, preferences and behavioral history. We employ the ubiquity of smartphones and the wealth of information available from social media to address the challenge of building proactive context-aware systems. We have implemented and evaluated a few approaches, including some as part of the Rover framework, to achieve the paradigm of Proactive Context-aware Computing. Rover is a context-aware research platform which has been evolving for the last 6 years. Since location is one of the most important context for users, we have developed ‘Locus’, an indoor localization, tracking and navigation system for multi-story buildings. Other important dimensions of users’ context include the activities that they are engaged in. To this end, we have developed ‘SenseMe’, a system that leverages the smartphone and its multiple sensors in order to perform multidimensional context and activity recognition for users. As part of the ‘SenseMe’ project, we also conducted an exploratory study of privacy, trust, risks and other concerns of users with smart phone based personal sensing systems and applications. To determine what information would be relevant to users’ situations, we have developed ‘TellMe’ - a system that employs a new, flexible and scalable approach based on Natural Language Processing techniques to perform bootstrapped discovery and ranking of relevant information in context-aware systems. In order to personalize the relevant information, we have also developed an algorithm and system for mining a broad range of users’ preferences from their social network profiles and activities. For recommending new information to the users based on their past behavior and context history (such as visited locations, activities and time), we have developed a recommender system and approach for performing multi-dimensional collaborative recommendations using tensor factorization. For timely delivery of personalized and relevant information, it is essential to anticipate and predict users’ behavior. To this end, we have developed a unified infrastructure, within the Rover framework, and implemented several novel approaches and algorithms that employ various contextual features and state of the art machine learning techniques for building diverse behavioral models of users. Examples of generated models include classifying users’ semantic places and mobility states, predicting their availability for accepting calls on smartphones and inferring their device charging behavior. Finally, to enable proactivity in context-aware systems, we have also developed a planning framework based on HTN planning. Together, these works provide a major push in the direction of proactive context-aware computing.

La comunicación de los diarios de Cuenca en las redes sociales

La presente tesis, titulada “La Comunicación de los Diarios de Cuenca en las redes sociales”, tuvo como finalidadidentificar y reconocer el punto actual en que se encuentra el uso de las redes sociales por parte de los Diarios cuencanos; para ello se realizó un estudio que permitió determinar el modelo comunicativo imperante. El proceso investigativo se basó en una observación cuantitativa y comparativa de las cuentas de los Diarios de Cuenca: EL TIEMPO, EL MERCURIO, LA TARDE y EL MORLACO DIGITAL, en Twitter, Facebook, YouTube y Flickr en los meses de agosto, septiembre y octubre de 2012. A través del seguimiento se obtuvo un registro referente a la interactividad, la influencia y el contenido de los diferentes mensajes publicados. Este análisis permitió concluir que los Diarios locales traspasan su modelo lineal, de sus versiones tradicionales, al sistema comunicativo de las redes sociales. Y, además que, a pesar de que esta forma de comunicación impuesta por los Diarios no encaja en la perspectiva de comunicación existente en las redes sociales, las cuentas de los diarios son influyentes, sobre todo en temas de relativa trascendencia informativa.

Datasets, features, learning, and models in visual recognition

Visual recognition is a fundamental research topic in computer vision. This dissertation explores datasets, features, learning, and models used for visual recognition. In order to train visual models and evaluate different recognition algorithms, this dissertation develops an approach to collect object image datasets on web pages using an analysis of text around the image and of image appearance. This method exploits established online knowledge resources (Wikipedia pages for text; Flickr and Caltech data sets for images). The resources provide rich text and object appearance information. This dissertation describes results on two datasets. The first is Berg’s collection of 10 animal categories; on this dataset, we significantly outperform previous approaches. On an additional set of 5 categories, experimental results show the effectiveness of the method. Images are represented as features for visual recognition. This dissertation introduces a text-based image feature and demonstrates that it consistently improves performance on hard object classification problems. The feature is built using an auxiliary dataset of images annotated with tags, downloaded from the Internet. Image tags are noisy. The method obtains the text features of an unannotated image from the tags of its k-nearest neighbors in this auxiliary collection. A visual classifier presented with an object viewed under novel circumstances (say, a new viewing direction) must rely on its visual examples. This text feature may not change, because the auxiliary dataset likely contains a similar picture. While the tags associated with images are noisy, they are more stable when appearance changes. The performance of this feature is tested using PASCAL VOC 2006 and 2007 datasets. This feature performs well; it consistently improves the performance of visual object classifiers, and is particularly effective when the training dataset is small. With more and more collected training data, computational cost becomes a bottleneck, especially when training sophisticated classifiers such as kernelized SVM. This dissertation proposes a fast training algorithm called Stochastic Intersection Kernel Machine (SIKMA). This proposed training method will be useful for many vision problems, as it can produce a kernel classifier that is more accurate than a linear classifier, and can be trained on tens of thousands of examples in two minutes. It processes training examples one by one in a sequence, so memory cost is no longer the bottleneck to process large scale datasets. This dissertation applies this approach to train classifiers of Flickr groups with many group training examples. The resulting Flickr group prediction scores can be used to measure image similarity between two images. Experimental results on the Corel dataset and a PASCAL VOC dataset show the learned Flickr features perform better on image matching, retrieval, and classification than conventional visual features. Visual models are usually trained to best separate positive and negative training examples. However, when recognizing a large number of object categories, there may not be enough training examples for most objects, due to the intrinsic long-tailed distribution of objects in the real world. This dissertation proposes an approach to use comparative object similarity. The key insight is that, given a set of object categories which are similar and a set of categories which are dissimilar, a good object model should respond more strongly to examples from similar categories than to examples from dissimilar categories. This dissertation develops a regularized kernel machine algorithm to use this category dependent similarity regularization. Experiments on hundreds of categories show that our method can make significant improvement for categories with few or even no positive examples.

The role of context in image annotation and recommendation

With the rise of smart phones, lifelogging devices (e.g. Google Glass) and popularity of image sharing websites (e.g. Flickr), users are capturing and sharing every aspect of their life online producing a wealth of visual content. Of these uploaded images, the majority are poorly annotated or exist in complete semantic isolation making the process of building retrieval systems difficult as one must firstly understand the meaning of an image in order to retrieve it. To alleviate this problem, many image sharing websites offer manual annotation tools which allow the user to “tag” their photos, however, these techniques are laborious and as a result have been poorly adopted; Sigurbjörnsson and van Zwol (2008) showed that 64% of images uploaded to Flickr are annotated with < 4 tags. Due to this, an entire body of research has focused on the automatic annotation of images (Hanbury, 2008; Smeulders et al., 2000; Zhang et al., 2012a) where one attempts to bridge the semantic gap between an image’s appearance and meaning e.g. the objects present. Despite two decades of research the semantic gap still largely exists and as a result automatic annotation models often offer unsatisfactory performance for industrial implementation. Further, these techniques can only annotate what they see, thus ignoring the “bigger picture” surrounding an image (e.g. its location, the event, the people present etc). Much work has therefore focused on building photo tag recommendation (PTR) methods which aid the user in the annotation process by suggesting tags related to those already present. These works have mainly focused on computing relationships between tags based on historical images e.g. that NY and timessquare co-exist in many images and are therefore highly correlated. However, tags are inherently noisy, sparse and ill-defined often resulting in poor PTR accuracy e.g. does NY refer to New York or New Year? This thesis proposes the exploitation of an image’s context which, unlike textual evidences, is always present, in order to alleviate this ambiguity in the tag recommendation process. Specifically we exploit the “what, who, where, when and how” of the image capture process in order to complement textual evidences in various photo tag recommendation and retrieval scenarios. In part II, we combine text, content-based (e.g. # of faces present) and contextual (e.g. day-of-the-week taken) signals for tag recommendation purposes, achieving up to a 75% improvement to precision@5 in comparison to a text-only TF-IDF baseline. We then consider external knowledge sources (i.e. Wikipedia & Twitter) as an alternative to (slower moving) Flickr in order to build recommendation models on, showing that similar accuracy could be achieved on these faster moving, yet entirely textual, datasets. In part II, we also highlight the merits of diversifying tag recommendation lists before discussing at length various problems with existing automatic image annotation and photo tag recommendation evaluation collections. In part III, we propose three new image retrieval scenarios, namely “visual event summarisation”, “image popularity prediction” and “lifelog summarisation”. In the first scenario, we attempt to produce a rank of relevant and diverse images for various news events by (i) removing irrelevant images such memes and visual duplicates (ii) before semantically clustering images based on the tweets in which they were originally posted. Using this approach, we were able to achieve over 50% precision for images in the top 5 ranks. In the second retrieval scenario, we show that by combining contextual and content-based features from images, we are able to predict if it will become “popular” (or not) with 74% accuracy, using an SVM classifier. Finally, in chapter 9 we employ blur detection and perceptual-hash clustering in order to remove noisy images from lifelogs, before combining visual and geo-temporal signals in order to capture a user’s “key moments” within their day. We believe that the results of this thesis show an important step towards building effective image retrieval models when there lacks sufficient textual content (i.e. a cold start).