Testing of Android testing tools: development of a benchmark for the evaluation

With the ever growing trend of smart phones and tablets, Android is becoming more and more popular everyday. With more than one billion active users i to date, Android is the leading technology in smart phone arena. In addition to that, Android also runs on Android TV, Android smart watches and cars. Therefore, in recent years, Android applications have become one of the major development sectors in software industry. As of mid 2013, the number of published applications on Google Play had exceeded one million and the cumulative number of downloads was more than 50 billionii. A 2013 survey also revealed that 71% of the mobile application developers work on developing Android applicationsiii. Considering this size of Android applications, it is quite evident that people rely on these applications on a daily basis for the completion of simple tasks like keeping track of weather to rather complex tasks like managing one’s bank accounts. Hence, like every other kind of code, Android code also needs to be verified in order to work properly and achieve a certain confidence level. Because of the gigantic size of the number of applications, it becomes really hard to manually test Android applications specially when it has to be verified for various versions of the OS and also, various device configurations such as different screen sizes and different hardware availability. Hence, recently there has been a lot of work on developing different testing methods for Android applications in Computer Science fraternity. The model of Android attracts researchers because of its open source nature. It makes the whole research model more streamlined when the code for both, application and the platform are readily available to analyze. And hence, there has been a great deal of research in testing and static analysis of Android applications. A great deal of this research has been focused on the input test generation for Android applications. Hence, there are a several testing tools available now, which focus on automatic generation of test cases for Android applications. These tools differ with one another on the basis of their strategies and heuristics used for this generation of test cases. But there is still very little work done on the comparison of these testing tools and the strategies they use. Recently, some research work has been carried outiv in this regard that compared the performance of various available tools with respect to their respective code coverage, fault detection, ability to work on multiple platforms and their ease of use. It was done, by running these tools on a total of 60 real world Android applications. The results of this research showed that although effective, these strategies being used by the tools, also face limitations and hence, have room for improvement. The purpose of this thesis is to extend this research into a more specific and attribute-­‐ oriented way. Attributes refer to the tasks that can be completed using the Android platform. It can be anything ranging from a basic system call for receiving an SMS to more complex tasks like sending the user to another application from the current one. The idea is to develop a benchmark for Android testing tools, which is based on the performance related to these attributes. This will allow the comparison of these tools with respect to these attributes. For example, if there is an application that plays some audio file, will the testing tool be able to generate a test input that will warrant the execution of this audio file? Using multiple applications using different attributes, it can be visualized that which testing tool is more useful for which kinds of attributes. In this thesis, it was decided that 9 attributes covering the basic nature of tasks, will be targeted for the assessment of three testing tools. Later this can be done for much more attributes to compare even more testing tools. The aim of this work is to show that this approach is effective and can be used on a much larger scale. One of the flagship features of this work, which also differentiates it with the previous work, is that the applications used, are all specially made for this research. The reason for doing that is to analyze just that specific attribute in isolation, which the application is focused on, and not allow the tool to get bottlenecked by something trivial, which is not the main attribute under testing. This means 9 applications, each focused on one specific attribute. The main contributions of this thesis are: A summary of the three existing testing tools and their respective techniques for automatic test input generation of Android Applications. • A detailed study of the usage of these testing tools using the 9 applications specially designed and developed for this study. • The analysis of the obtained results of the study carried out. And a comparison of the performance of the selected tools.

Study of OLSR protocol with Network Simulator 2 over Linux

Until a few years ago, most of the network communications were based in the wire as the physical media, but due to the advances and the maturity of the wireless communications, this is changing. Nowadays wireless communications offers fast, secure, efficient and reliable connections. Mobile communications are in expansion, clearly driven by the use of smart phones and other mobile devices, the use of laptops, etc… Besides that point, the inversion in the installation and maintenance of the physical medium is much lower than in wired communications, not only because the air has no cost, but because the installation and maintenance of the wire require a high economic cost. Besides the economic cost we find that wire is a more vulnerable medium to external threats such as noise, sabotages, etc… There are two different types of wireless networks: those which the structure is part of the network itself and those which have a lack of structure or any centralization, in a way that the devices that form part of the network can connect themselves in a dynamic and random way, handling also the routing of every control and information messages, this kind of networks is known as Ad-hoc. In the present work we will proceed to study one of the multiple wireless protocols that allows mobile communications, it is Optimized Link State Routing, from now on, OLSR, it is an pro-active routing, standard mechanism that works in a distributed in order to stablish the connections among the different nodes that belong to a wireless network. Thanks to this protocol it is possible to get all the routing tables in all the devices correctly updated every moment through the periodical transmission of control messages and on this way allow a complete connectivity among the devices that are part of the network and also, allow access to other external networks such as virtual private networks o Internet. This protocol could be perfectly used in environments such as airports, malls, etc… The update of the routing tables in all the devices is got thanks to the periodical transmission of control messages and finally it will offer connectivity among all the devices and the corresponding external networks. For the study of OLSR protocol we will have the help of the network simulator “Network Simulator 2”, a freeware network simulator programmed in C++ based in discrete events. This simulator is used mainly in educational and research environments and allows a very extensive range of protocols, both, wired networks protocols and wireless network protocols, what is going to be really useful to proceed to the simulation of different configurations of networks and protocols. In the present work we will also study different simulations with Network Simulator 2, in different scenarios with different configurations, wired networks, and Ad-hoc networks, where we will study OLSR Protocol. RESUMEN. Hasta hace pocos años, la mayoría de las comunicaciones de red estaban basadas en el cable como medio físico pero debido al avance y madurez alcanzados en el campo de las comunicaciones inalámbricas esto está cambiando. Hoy día las comunicaciones inalámbricas nos ofrecen conexiones veloces, seguras, eficientes y fiables. Las comunicaciones móviles se encuentran en su momento de máxima expansión, claramente impulsadas por el uso de teléfonos y demás dispositivos móviles, el uso de portátiles, etc… Además la inversión a realizar en la instalación y el mantenimiento del medio físico en las comunicaciones móviles es muchísimo menor que en comunicaciones por cable, ya no sólo porque el aire no tenga coste alguno, sino porque la instalación y mantenimiento del cable precisan de un elevado coste económico por norma. Además del coste económico nos encontramos con que es un medio más vulnerable a amenazas externas tales como el ruido, escuchas no autorizadas, sabotajes, etc… Existen dos tipos de redes inalámbricas: las constituidas por una infraestructura que forma parte más o menos de la misma y las que carecen de estructura o centralización alguna, de modo que los dispositivos que forman parte de ella pueden conectarse de manera dinámica y arbitraria entre ellos, encargándose además del encaminamiento de todos los mensajes de control e información, a este tipo de redes se las conoce como redes Ad-hoc. En el presente Proyecto de Fin de Carrera se procederá al estudio de uno de los múltiples protocolos inalámbricos que permiten comunicaciones móviles, se trata del protocolo inalámbrico Optimized Link State Routing, de ahora en adelante OLSR, un mecanismo estándar de enrutamiento pro-activo, que trabaja de manera distribuida para establecer las conexiones entre los nodos que formen parte de las redes inalámbricas Ad-hoc, las cuales carecen de un nodo central y de una infraestructura pre-existente. Gracias a este protocolo es posible conseguir que todos los equipos mantengan en todo momento las tablas de ruta actualizadas correctamente mediante la transmisión periódica de mensajes de control y así permitir una completa conectividad entre todos los equipos que formen parte de la red y, a su vez, también permitir el acceso a otras redes externas tales como redes privadas virtuales o Internet. Este protocolo sería usado en entornos tales como aeropuertos La actualización de las tablas de enrutamiento de todos los equipos se conseguirá mediante la transmisión periódica de mensajes de control y así finalmente se podrá permitir conectividad entre todos los equipos y con las correspondientes redes externas. Para el estudio del protocolo OLSR contaremos con el simulador de redes Network Simulator 2, un simulador de redes freeware programado en C++ basado en eventos discretos. Este simulador es usado principalmente en ambientes educativos y de investigación y permite la simulación tanto de protocolos unicast como multicast. El campo donde más se utiliza es precisamente en el de la investigación de redes móviles Ad-hoc. El simulador Network Simulator 2 no sólo implementa el protocolo OLSR, sino que éste implementa una amplia gama de protocolos, tanto de redes cableadas como de redes inalámbricas, lo cual va a sernos de gran utilidad para proceder a la simulación de distintas configuraciones de redes y protocolos. En el presente Proyecto de Fin de Carrera se estudiarán también diversas simulaciones con el simulador NS2 en diferentes escenarios con diversas configuraciones; redes cableadas, redes inalámbricas Ad-hoc, donde se estudiará el protocolo antes mencionado: OLSR. Este Proyecto de Fin de Carrera consta de cuatro apartados distintos: Primeramente se realizará el estudio completo del protocolo OLSR, se verán los beneficios y contrapartidas que ofrece este protocolo inalámbrico. También se verán los distintos tipos de mensajes existentes en este protocolo y unos pequeños ejemplos del funcionamiento del protocolo OLSR. Seguidamente se hará una pequeña introducción al simulador de redes Network Simulator 2, veremos la historia de este simulador, y también se hará referencia a la herramienta extra NAM, la cual nos permitirá visualizar el intercambio de paquetes que se produce entre los diferentes dispositivos de nuestras simulaciones de forma intuitiva y amigable. Se hará mención a la plataforma MASIMUM, encargada de facilitar en un entorno académico software y documentación a sus alumnos con el fin de facilitarles la investigación y la simulación de redes y sensores Ad-hoc. Finalmente se verán dos ejemplos, uno en el que se realizará una simulación entre dos PCs en un entorno Ethernet y otro ejemplo en el que se realizará una simulación inalámbrica entre cinco dispositivos móviles mediante el protocolo a estudiar, OLSR.

E-government strategy in Turkey:a case for m-government

Market orientation strategies are now expected to be integrated and enacted by firms and governments alike. While private services will surely continue to take the lead in mobile strategy orientation, others such as government and Non-Governmental Organizations (NGOs) are also becoming prominent Mobile Players (m-Players). Enhanced data services through smart phones are raising expectations that governments will finally deliver services that are in line with a consumer ICT lifestyle. To date, it is not certain which form of technological standards will take the lead, e.g. enhanced m-services or traditional Internet-based applications. Yet, with the introduction of interactive applications and fully transactional services via 3G smart phones, the currently untapped segment of the population (without computers) have the potential to gain access to government services at a low cost.

Detecting and monitoring the symptoms of Parkinson's disease using smartphones:a pilot study

Background: Remote, non-invasive and objective tests that can be used to support expert diagnosis for Parkinson's disease (PD) are lacking. Methods: Participants underwent baseline in-clinic assessments, including the Unified Parkinson's Disease Rating Scale (UPDRS), and were provided smartphones with an Android operating system that contained a smartphone application that assessed voice, posture, gait, finger tapping, and response time. Participants then took the smart phones home to perform the five tasks four times a day for a month. Once a week participants had a remote (telemedicine) visit with a Parkinson disease specialist in which a modified (excluding assessments of rigidity and balance) UPDRS performed. Using statistical analyses of the five tasks recorded using the smartphone from 10 individuals with PD and 10 controls, we sought to: (1) discriminate whether the participant had PD and (2) predict the modified motor portion of the UPDRS. Results: Twenty participants performed an average of 2.7 tests per day (68.9% adherence) for the study duration (average of 34.4 days) in a home and community setting. The analyses of the five tasks differed between those with Parkinson disease and those without. In discriminating participants with PD from controls, the mean sensitivity was 96.2% (SD 2%) and mean specificity was 96.9% (SD 1.9%). The mean error in predicting the modified motor component of the UPDRS (range 11-34) was 1.26 UPDRS points (SD 0.16). Conclusion: Measuring PD symptoms via a smartphone is feasible and has potential value as a diagnostic support tool.

Mobile Access to Cultural and Historical Heritage Web Pages

The main focus of attention in this paper is the access to websites of leading regional, national and world museums. Their accessibility from desktop and mobile devices is hereby evaluated. The paper covers a detailed analysis of the mobile operating systems, their market share in Bulgaria and the opportunities they offer to the users. The need for contemporary cultural and historical heritage websites which maintain lighter but fully-functioning mobile versions, facilitating and making their access from portable devices easier, is also well-grounded here.

High-accuracy discrimination of Parkinson’s Disease participants from healthy controls using smartphones

Objective: To test the practicality and effectiveness of cheap, ubiquitous, consumer-grade smartphones to discriminate Parkinson’s disease (PD) subjects from healthy controls, using self-administered tests of gait and postural sway. Background: Existing tests for the diagnosis of PD are based on subjective neurological examinations, performed in-clinic. Objective movement symptom severity data, collected using widely-accessible technologies such as smartphones, would enable the remote characterization of PD symptoms based on self-administered, behavioral tests. Smartphones, when backed up by interviews using web-based videoconferencing, could make it feasible for expert neurologists to perform diagnostic testing on large numbers of individuals at low cost. However, to date, the compliance rate of testing using smart-phones has not been assessed. Methods: We conducted a one-month controlled study with twenty participants, comprising 10 PD subjects and 10 controls. All participants were provided identical LG Optimus S smartphones, capable of recording tri-axial acceleration. Using these smartphones, patients conducted self-administered, short (less than 5 minute) controlled gait and postural sway tests. We analyzed a wide range of summary measures of gait and postural sway from the accelerometry data. Using statistical machine learning techniques, we identified discriminating patterns in the summary measures in order to distinguish PD subjects from controls. Results: Compliance was high all 20 participants performed an average of 3.1 tests per day for the duration of the study. Using this test data, we demonstrated cross-validated sensitivity of 98% and specificity of 98% in discriminating PD subjects from healthy controls. Conclusions: Using consumer-grade smartphone accelerometers, it is possible to distinguish PD from healthy controls with high accuracy. Since these smartphones are inexpensive (around $30 each) and easily available, and the tests are highly non-invasive and objective, we envisage that this kind of smartphone-based testing could radically increase the reach and effectiveness of experts in diagnosing PD.

Emerging perspectives on the design, use, and evaluation of mobile and handheld devices

Human-computer interaction is a growing field of study in which researchers and professionals aim to understand and evaluate the impact of new technologies on human behavior. With the integration of smart phones, tablets, and other portable devices into everyday life, there is a greater need to understand the influence of such technology on the human experience. Emerging Perspectives on the Design, Use, and Evaluation of Mobile and Handheld Devices is an authoritative reference source consisting of the latest scholarly research and theories from international experts and professionals on the topic of human-computer interaction with mobile devices. Featuring a comprehensive collection of chapters on critical topics in this dynamic field, this publication is an essential reference source for researchers, educators, students, and practitioners interested in the use of mobile and handheld devices and their impact on individuals and society as a whole. This publication features timely, research-based chapters pertaining to topics in the design and evaluation of smart devices including, but not limited to, app stores, category-based interfaces, gamified mobility applications, mobile interaction, mobile learning, pervasive multimodal applications, smartphone interaction, and social media use.

WiFi and LTE Coexistence in the Unlicensed Spectrum

Today, smart-phones have revolutionized wireless communication industry towards an era of mobile data. To cater for the ever increasing data traffic demand, it is of utmost importance to have more spectrum resources whereby sharing under-utilized spectrum bands is an effective solution. In particular, the 4G broadband Long Term Evolution (LTE) technology and its foreseen 5G successor will benefit immensely if their operation can be extended to the under-utilized unlicensed spectrum. In this thesis, first we analyze WiFi 802.11n and LTE coexistence performance in the unlicensed spectrum considering multi-layer cell layouts through system level simulations. We consider a time division duplexing (TDD)-LTE system with an FTP traffic model for performance evaluation. Simulation results show that WiFi performance is more vulnerable to LTE interference, while LTE performance is degraded only slightly. Based on the initial findings, we propose a Q-Learning based dynamic duty cycle selection technique for configuring LTE transmission gaps, so that a satisfactory throughput is maintained both for LTE and WiFi systems. Simulation results show that the proposed approach can enhance the overall capacity performance by 19% and WiFi capacity performance by 77%, hence enabling effective coexistence of LTE and WiFi systems in the unlicensed band.

IF-MANET: Interoperable framework for heterogeneous mobile ad hoc networks

The advances in low power micro-processors, wireless networks and embedded systems have raised the need to utilize the significant resources of mobile devices. These devices for example, smart phones, tablets, laptops, wearables, and sensors are gaining enormous processing power, storage capacity and wireless bandwidth. In addition, the advancement in wireless mobile technology has created a new communication paradigm via which a wireless network can be created without any priori infrastructure called mobile ad hoc network (MANET). While progress is being made towards improving the efficiencies of mobile devices and reliability of wireless mobile networks, the mobile technology is continuously facing the challenges of un-predictable disconnections, dynamic mobility and the heterogeneity of routing protocols. Hence, the traditional wired, wireless routing protocols are not suitable for MANET due to its unique dynamic ad hoc nature. Due to the reason, the research community has developed and is busy developing protocols for routing in MANET to cope with the challenges of MANET. However, there are no single generic ad hoc routing protocols available so far, which can address all the basic challenges of MANET as mentioned before. Thus this diverse range of ever growing routing protocols has created barriers for mobile nodes of different MANET taxonomies to intercommunicate and hence wasting a huge amount of valuable resources. To provide interaction between heterogeneous MANETs, the routing protocols require conversion of packets, meta-model and their behavioural capabilities. Here, the fundamental challenge is to understand the packet level message format, meta-model and behaviour of different routing protocols, which are significantly different for different MANET Taxonomies. To overcome the above mentioned issues, this thesis proposes an Interoperable Framework for heterogeneous MANETs called IF-MANET. The framework hides the complexities of heterogeneous routing protocols and provides a homogeneous layer for seamless communication between these routing protocols. The framework creates a unique Ontology for MANET routing protocols and a Message Translator to semantically compare the packets and generates the missing fields using the rules defined in the Ontology. Hence, the translation between an existing as well as newly arriving routing protocols will be achieved dynamically and on-the-fly. To discover a route for the delivery of packets across heterogeneous MANET taxonomies, the IF-MANET creates a special Gateway node to provide cluster based inter-domain routing. The IF-MANET framework can be used to develop different middleware applications. For example: Mobile grid computing that could potentially utilise huge amounts of aggregated data collected from heterogeneous mobile devices. Disaster & crises management applications can be created to provide on-the-fly infrastructure-less emergency communication across organisations by utilising different MANET taxonomies.

Acceleration and visualization of Dynamic Network Optimization

With the emerging prevalence of smart phones and 4G LTE networks, the demand for faster-better-cheaper mobile services anytime and anywhere is ever growing. The Dynamic Network Optimization (DNO) concept emerged as a solution that optimally and continuously tunes the network settings, in response to varying network conditions and subscriber needs. Yet, the DNO realization is still at infancy, largely hindered by the bottleneck of the lengthy optimization runtime. This paper presents the design and prototype of a novel cloud based parallel solution that further enhances the scalability of our prior work on various parallel solutions that accelerate network optimization algorithms. The solution aims to satisfy the high performance required by DNO, preliminarily on a sub-hourly basis. The paper subsequently visualizes a design and a full cycle of a DNO system. A set of potential solutions to large network and real-time DNO are also proposed. Overall, this work creates a breakthrough towards the realization of DNO.

Myanmar MIL Workshop Series

The Mobile Information Literacy curriculum is a growing collection of training materials designed to build literacies for the millions of people worldwide coming online every month via a mobile phone. Most information information and digital literacy curricula were designed for a PC age, and public and private organizations around the world have used these curricula to help newcomers use computers and the internet effectively and safely. The better curricula address not only skills, but also concepts and attitudes. The central question for this project is: what are the relevant skills, concepts, and attitudes for people using mobiles, not PCs, to access the internet? As part of the Information Strategies for Societies in Transition project, we developed a six-module curriculum for mobile-first users. The project is situated in Myanmar, a country undergoing massive political, economic, and social changes, and where mobile penetration is expected to reach 80% by the end of 2015 from just 4% in 2014. Combined with the country’s history of media censorship, Myanmar presents unique challenges for addressing the needs of people who need the ability to find and evaluate the quality and credibility of information obtained online, understand how to create and share online information effectively, and participate safely and securely.

Myanmar DIL WORKSHOP SERIES Unit 4: Working Online and Using Information

The Mobile Information Literacy curriculum is a growing collection of training materials designed to build literacies for the millions of people worldwide coming online every month via a mobile phone. Most information information and digital literacy curricula were designed for a PC age, and public and private organizations around the world have used these curricula to help newcomers use computers and the internet effectively and safely. The better curricula address not only skills, but also concepts and attitudes. The central question for this project is: what are the relevant skills, concepts, and attitudes for people using mobiles, not PCs, to access the internet? As part of the Information Strategies for Societies in Transition project, we developed a six-module curriculum for mobile-first users. The project is situated in Myanmar, a country undergoing massive political, economic, and social changes, and where mobile penetration is expected to reach 80% by the end of 2015 from just 4% in 2014. Combined with the country’s history of media censorship, Myanmar presents unique challenges for addressing the needs of people who need the ability to find and evaluate the quality and credibility of information obtained online, understand how to create and share online information effectively, and participate safely and securely.

Context-awareness privacy in data communications

Internet users consume online targeted advertising based on information collected about them and voluntarily share personal information in social networks. Sensor information and data from smart-phones is collected and used by applications, sometimes in unclear ways. As it happens today with smartphones, in the near future sensors will be shipped in all types of connected devices, enabling ubiquitous information gathering from the physical environment, enabling the vision of Ambient Intelligence. The value of gathered data, if not obvious, can be harnessed through data mining techniques and put to use by enabling personalized and tailored services as well as business intelligence practices, fueling the digital economy. However, the ever-expanding information gathering and use undermines the privacy conceptions of the past. Natural social practices of managing privacy in daily relations are overridden by socially-awkward communication tools, service providers struggle with security issues resulting in harmful data leaks, governments use mass surveillance techniques, the incentives of the digital economy threaten consumer privacy, and the advancement of consumergrade data-gathering technology enables new inter-personal abuses. A wide range of fields attempts to address technology-related privacy problems, however they vary immensely in terms of assumptions, scope and approach. Privacy of future use cases is typically handled vertically, instead of building upon previous work that can be re-contextualized, while current privacy problems are typically addressed per type in a more focused way. Because significant effort was required to make sense of the relations and structure of privacy-related work, this thesis attempts to transmit a structured view of it. It is multi-disciplinary - from cryptography to economics, including distributed systems and information theory - and addresses privacy issues of different natures. As existing work is framed and discussed, the contributions to the state-of-theart done in the scope of this thesis are presented. The contributions add to five distinct areas: 1) identity in distributed systems; 2) future context-aware services; 3) event-based context management; 4) low-latency information flow control; 5) high-dimensional dataset anonymity. Finally, having laid out such landscape of the privacy-preserving work, the current and future privacy challenges are discussed, considering not only technical but also socio-economic perspectives.

Architecture for IoT Node and Platform

Only recently, during the past five years, consumer electronics has been evolving rapidly. Many products have started to include “smart home” capabilities, enabling communication and interoperability of various smart devices. Even more devices and sensors can be remote controlled and monitored through cloud services. While the smart home systems have become very affordable to average consumer compared to the early solutions decades ago, there are still many issues and things that need to be fixed or improved upon: energy efficiency, connectivity with other devices and applications, security and privacy concerns, reliability, and response time. This paper focuses on designing Internet of Things (IoT) node and platform architectures that take these issues into account, notes other currently used solutions, and selects technologies in order to provide better solution. The node architecture aims for energy efficiency and modularity, while the platform architecture goals are in scalability, portability, maintainability, performance, and modularity. Moreover, the platform architecture attempts to improve user experience by providing higher reliability and lower response time compared to the alternative platforms. The architectures were developed iteratively using a development process involving research, planning, design, implementation, testing, and analysis. Additionally, they were documented using Kruchten’s 4+1 view model, which is used to describe the use cases and different views of the architectures. The node architecture consisted of energy efficient hardware, FC3180 microprocessor and CC2520 RF transceiver, modular operating system, Contiki, and a communication protocol, AllJoyn, used for providing better interoperability with other IoT devices and applications. The platform architecture provided reliable low response time control, monitoring, and initial setup capabilities by utilizing web technologies on various devices such as smart phones, tablets, and computers. Furthermore, an optional cloud service was provided in order to control devices and monitor sensors remotely by utilizing scalable high performance technologies in the backend enabling low response time and high reliability.

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).