Automated classification of complex social survey questionnaires is crucial for large-scale social science research but faces significant reliability challenges due to intricate hierarchical label structures, severe class imbalance, semantic ambiguity, and incomplete data coverage. Conventional classification methods often struggle with these combined complexities, yielding results that lack trustworthiness. We introduce HOCM, a framework designed for trustworthy classification in complex, real-world taxonomies. It features two synergistic components: (1) memory-enhanced contrastive learning, tailored to learn robust representations from noisy, imbalanced data by leveraging quality-aware category memory banks; and (2) hierarchical uncertainty calibration, which enforces taxonomic consistency while providing reliable confidence estimates and identifying inputs falling outside well-represented known categories. Our evaluation on a large-scale, real-world social survey dataset—a challenging exemplar of our target problem class—demonstrates that HOCM maintains strong accuracy on known classes while effectively identifying uncertain cases, significantly boosting accuracy on confident predictions. Furthermore, it adeptly detects low-resource/unknown categories. HOCM provides a more reliable automated classification tool, enabling efficient expert review and enhancing the trustworthiness of analysis in domains with complex, hierarchical data.

Vocabularies such as the European Language Social Science Thesaurus (ELSST) and the CLOSER ontology are the foundational taxonomies capturing core social science concepts that form the foundations of large-scale longitudinal social science surveys. However, standard text embeddings often fail to capture the complex hierarchical and relational structures of the sociological concepts, relying on surface similarity. In this work, we propose a framework to model these nuances by adapting a large language model based text embedding model with a learnable diagonal Riemannian metric. This metric allows for a flexible geometry where dimensions can be scaled to reflect semantic importance. Additionally, we introduce a Hierarchical Ranking Loss with dynamic margins as the sole training objective to enforce the multi-level hierarchical constraints (e.g., distinguishing 'self' from narrower, broader, or related concepts, and all from 'unrelated' ones) from ELSST within the Riemannian space, such as ensuring a specific concept like 'social stratification' is correctly positioned by, for instance, being embedded closer to 'social inequality' (as its broader, related concept) and substantially further from an 'unrelated' concept like 'particle physics'. Lastly, we show that our parameter-efficient approach significantly outperforms strong contrastive learning and hyperbolic embedding baselines on hierarchical concept retrieval and classification tasks using the ELSST and CLOSER datasets. Visualizations confirm the learned embedding space exhibits a clear hierarchical structure. Our work offers a more accurate and geometrically informed method for representing complex sociological constructs.

Understanding online conversations has attracted research attention with the growth of social networks and online discussion forums. Content analysis of posts and replies in online conversations is difficult because each individual utterance is usually short and may implicitly refer to other posts within the same conversation. Thus, understanding individual posts requires capturing the conversational context and dependencies between different parts of a conversation tree and then encoding the context dependencies between posts and comments/replies into the language model. To this end, we propose a general-purpose mechanism to discover appropriate conversational context for various aspects about an online post in a conversation, such as whether it is informative, insightful, interesting or funny. Specifically, we design two families of Conversation Kernels, which explore different parts of the neighborhood of a post in the tree representing the conversation and through this, build relevant conversational context that is appropriate for each task being considered. We apply our developed method to conversations crawled from slashdot.org, which allows users to apply highly different labels to posts, such as `insightful', `funny', etc., and therefore provides an ideal experimental platform to study whether a framework such as Conversation Kernels is general-purpose and flexible enough to be adapted to disparately different conversation understanding tasks. We perform extensive experiments and find that context-augmented conversation kernels can significantly outperform transformer-based baselines, with absolute improvements in accuracy up to 20% and up to 19% for macro-F1 score. Our evaluations also show that conversation kernels outperform state-of-the-art large language models including GPT-4. We also showcase the generalizability and demonstrate that conversation kernels can be a general-purpose approach that flexibly handles distinctly different conversation understanding tasks in a unified manner.

No abstract available.

No abstract available.

Content Delivery Networks (CDNs) have been pivotal in the dramatic evolution of the Internet, handling the majority of data traffic for billions of connected users. Low-Earth-Orbit (LEO) satellite networks, such as Starlink, aim to revolutionize global connectivity by providing high-speed, low-latency Internet to remote regions. However, LEO satellite networks (LSNs) face challenges integrating with traditional CDNs, which rely on geographical proximity for efficient content delivery - a method that clashes with the operational dynamics of LSNs. In this paper, we scrutinize the operation of CDNs in the context of LSNs, using Starlink as a case study. We develop a browser extension NetMet that performs extensive web browsing experiments from controlled nodes using both Starlink and terrestrial Internet access. Additionally, we analyse crowdsourced speed tests from Starlink users to Cloudflare CDN servers globally. Our results indicate significant performance issues for Starlink users, stemming from the misalignment between terrestrial and satellite infrastructures. We then investigate the potential for SpaceCDNs which integrate CDN infrastructure directly within the LSNs, and show that this approach offers a promising alternative that decreases latencies by over 50%, making them comparable with the CDN experience of users behind terrestrial ISPs. Our aim is to stimulate further research and discussion on overcoming the challenges of effective content delivery with growing LSN offerings.

The recent development of decentralised and interoperable social networks (such as the "fediverse") creates new challenges for content moderators. This is because millions of posts generated on one server can easily "spread" to another, even if the recipient server has very different moderation policies. An obvious solution would be to leverage moderation tools to automatically tag (and filter) posts that contravene moderation policies, e.g. related to toxic speech. Recent work has exploited the conversational context of a post to improve this automatic tagging, e.g. using the replies to a post to help classify if it contains toxic speech. This has shown particular potential in environments with large training sets that contain complete conversations. This, however, creates challenges in a decentralised context, as a single conversation may be fragmented across multiple servers. Thus, each server only has a partial view of an entire conversation because conversations are often federated across servers in a non-synchronized fashion. To address this, we propose a decentralised conversation-aware content moderation approach suitable for the fediverse. Our approach employs a graph deep learning model (GraphNLI) trained locally on each server. The model exploits local data to train a model that combines post and conversational information captured through random walks to detect toxicity. We evaluate our approach with data from Pleroma, a major decentralised and interoperable micro-blogging network containing 2 million conversations. Our model effectively detects toxicity on larger instances, exclusively trained using their local post information (0.8837 macro-F1). Yet, we show that this approach does not perform well on smaller instances that do not possess sufficient local training data. Thus, in cases where a server contains insufficient data, we strategically retrieve information (posts or model parameters) from other servers to reconstruct larger conversations and improve results. With this, we show that we can attain a macro-F1 of 0.8826. Our approach has considerable scope to improve moderation in decentralised and interoperable social networks such as Pleroma or Mastodon.

For the viability of future “mission-critical” applications such as remote surgery and connected cars, the customers must trust the network connection and operators must adhere to the Service Level Agreement (SLA). The key to enabling trust between the customer and the operator is the transparency and accountability of the SLA. That is, the operators ensure the transparent and appropriate execution of SLA clauses (e.g., Quality of Service (QoS) and compensations if the SLA is violated). In this work, we argue that today’s network is highly volatile. Therefore, it is convenient for the operators to provide service guarantees for short-term rather than traditional long-term methods. Consequently, we advocate short-term and dynamic service contracts considering spatial and temporal characteristics rather than typical long-term contracts. We propose a Distributed Ledger Technology (DLT)-focused end-to-end transparent, accountable and automated resource provisioning system architecture in which are installed as smart contracts. In our architecture, resources can be requested and allocated dynamically and automatically with Quality-of-Service (QoS) monitoring. Our architecture is scalable through a side-channel based QoS monitoring protocol that guarantees data integrity and minimises the Permissioned Distributed Ledgers (PDL) updates. To measure the viability of our proposal, we first evaluate resource provisioning in the context of network slicing. Then we assess the DLT performance for smart contract execution, in both permissioned and permission-less settings. In the end, we evaluate the monitoring tools and compare and contrast sketches and bloomfilters, and justify our choice of bloomfilters.

Online forums that allow for participatory engagement between users have been transformative for the public discussion of many important issues. However, such conversations can sometimes escalate into full-blown exchanges of hate and misinformation. Existing approaches in natural language processing (NLP), such as deep learning models for classification tasks, use as inputs only a single comment or a pair of comments depending upon whether the task concerns the inference of properties of the individual comments or the replies between pairs of comments, respectively. However, in online conversations, comments and replies may be based on external context beyond the immediately relevant information that is input to the model. Therefore, being aware of the conversations’ surrounding contexts should improve the model’s performance for the inference task at hand. We propose GraphNLI , 1 a novel graph-based deep learning architecture that uses graph walks to incorporate the wider context of a conversation in a principled manner. Specifically, a graph walk starts from a given comment and samples “nearby” comments in the same or parallel conversation threads, which results in additional embeddings that are aggregated together with the initial comment’s embedding. We then use these enriched embeddings for downstream NLP prediction tasks that are important for online conversations. We evaluate GraphNLI on two such tasks - polarity prediction and misogynistic hate speech detection - and find that our model consistently outperforms all relevant baselines for both tasks. Specifically, GraphNLI with a biased root-seeking random walk performs with a macro- F 1 score of 3 and 6 percentage points better than the best-performing BERT-based baselines for the polarity prediction and hate speech detection tasks, respectively. We also perform extensive ablative experiments and hyperparameter searches to understand the efficacy of GraphNLI. This demonstrates the potential of context-aware models to capture the global context along with the local context of online conversations for these two tasks.

The introduction of ChatGPT and the subsequent improvement of Large Language Models (LLMs) have prompted more and more individuals to turn to the use of ChatBots, both for information and assistance with decision-making. However, the information the user is after is often not formulated by these ChatBots objectively enough to be provided with a definite, globally accepted answer.

Supervised machine learning approaches often rely on a "ground truth" label. However, obtaining one label through majority voting ignores the important subjectivity information in tasks such hate speech detection. Existing neural network models principally regard labels as categorical variables, while ignoring the semantic information in diverse label texts. In this paper, we propose AnnoBERT, a first-of-its-kind architecture integrating annotator characteristics and label text with a transformer-based model to detect hate speech, with unique representations based on each annotator's characteristics via Collaborative Topic Regression (CTR) and integrate label text to enrich textual representations. During training, the model associates annotators with their label choices given a piece of text; during evaluation, when label information is not available, the model predicts the aggregated label given by the participating annotators by utilising the learnt association. The proposed approach displayed an advantage in detecting hate speech, especially in the minority class and edge cases with annotator disagreement. Improvement in the overall performance is the largest when the dataset is more label-imbalanced, suggesting its practical value in identifying real-world hate speech, as the volume of hate speech in-the-wild is extremely small on social media, when compared with normal (non-hate) speech. Through ablation studies, we show the relative contributions of annotator embeddings and label text to the model performance, and tested a range of alternative annotator embeddings and label text combinations.

The acquisition of Twitter by Elon Musk has spurred controversy and uncertainty among Twitter users. The move raised both praise and concerns, particularly regarding Musk's views on free speech. As a result, a large number of Twitter users have looked for alternatives to Twitter. Mastodon, a decentralized micro-blogging social network, has attracted the attention of many users and the general media. In this paper, we analyze the migration of 136,009 users from Twitter to Mastodon. We inspect the impact that this has on the wider Mastodon ecosystem, particularly in terms of user-driven pressure towards centralization. We further explore factors that influence users to migrate, highlighting the effect of users' social networks. Finally, we inspect the behavior of individual users, showing how they utilize both Twitter and Mastodon in parallel. We find a clear difference in the topics discussed on the two platforms. This leads us to build classifiers to explore if migration is predictable. Through feature analysis, we find that the content of tweets as well as the number of URLs, the number of likes, and the length of tweets are effective metrics for the prediction of user migration.

With the advent of 5G+ services, it has become increasingly convenient for mobile users to enjoy high-quality multimedia content from CDN driven streaming and catch-up TV services (Netflix, iPlayer) in the (post-) COVID over-the-top (OTT) content rush. To relieve ISP owned fixed-line networks from CDN streamed multimedia traffic, system ideas (e.g., Wi-Stitch in [ 45 ]) have been proposed to (a) leverage 5G services and enable consumers to share cached multimedia content at the edge, and (b) consequently, and more importantly, reduce IP traffic at the core network. Unfortunately, given that contemporary multimedia content might be a monetized asset, these ideas do not take this important fact into account for shared content. We present EdgeMart —a content provider (CP) federated, and computationally sustainable networked (graphical) market economy for paid-sharing of cached licensed (OTT) content with autonomous users of a wireless edge network (WEN). EdgeMart is a unique oligopoly multimedia market (economy) that comprises competing networked sub-markets of non-cooperative content sellers/buyers—each sub-market consisting of a single buyer connected (networked) to only a subset of sellers. We prove that for any WEN-supported supply-demand topology , a pure strategy EdgeMart equilibrium exists that is (a) nearly efficient (in a microeconomic sense) indicating economy sustainability, (b) robust to edge user entry/exit, and (c) can be reached in poly-time (indicating computational sustainability). In addition, we experimentally show that for physical WENs of varying densities, a rationally selfish EdgeMart economy induces similar orders of multimedia IP traffic savings when compared to the ideal (relatively less practical), altruistic , and non-monetized “economy” implemented atop the recently introduced Wi-Stitch WEN-based content trading architecture. Moreover, the EdgeMart concept helps envision a regulated edge economy of opportunistic (pay per licensed file) client services for commercial OTT platforms.

LEO satellite "mega-constellations" such as SpaceX's Starlink, Amazon's Kuiper, OneWeb are launching thousands of satellites annually, promising high-bandwidth low-latency connectivity. To quantify the achievable performance of such providers, we carry out a measurement study of the spatial and temporal characteristics as well as the geographic variability of the connectivity provided by Starlink, the current leader in this space. We do this by building and deploying a browser extension that provides data about web performance seen by 28 users from 10 cities across the world. We complement this with performance tests run from three measurement nodes hosted by volunteer enthusiasts in the UK, EU and USA. Our findings suggest that although Starlink offers some of the best web performance figures among the ISPs observed, there are important sources of variability in performance such as weather conditions. The bent-pipe connection to a satellite and back to earth also forms a significant component of the observed latency. We also observe frequent and significant packet losses of up to 50% of packets, which appear to be correlated with handovers between satellites. This has an effect on achievable throughput even when using modern congestion control protocols such as BBR or CUBIC.

Online forums that allow participatory engagement between users have been transformative for public discussion of important issues. However, debates on such forums can sometimes escalate into full blown exchanges of hate or misinformation. An important tool in understanding and tackling such problems is to be able to infer the argumentative relation of whether a reply is supporting or attacking the post it is replying to. This so called polarity prediction task is difficult because replies may be based on external context beyond a post and the reply whose polarity is being predicted. We propose GraphNLI, a novel graph-based deep learning architecture that uses graph walk techniques to capture the wider context of a discussion thread in a principled fashion. Specifically, we propose methods to perform root-seeking graph walks that start from a post and captures its surrounding context to generate additional embeddings for the post. We then use these embeddings to predict the polarity relation between a reply and the post it is replying to. We evaluate the performance of our models on a curated debate dataset from Kialo, an online debating platform. Our model outperforms relevant baselines, including S-BERT, with an overall accuracy of 83%.

WhatsApp is a popular messaging app used by over a billion users around the globe. Due to this popularity, understanding misbehavior on WhatsApp is an important issue. The sending of unwanted junk messages by unknown contacts via WhatsApp remains understudied by researchers, in part because of the end-to-end encryption offered by the platform. We address this gap by studying junk messaging on a multilingual dataset of 2.6M messages sent to 5K public WhatsApp groups in India. We characterise both junk content and senders. We find that nearly 1 in 10 messages is unwanted content sent by junk senders, and a number of unique strategies are employed to reflect challenges faced on WhatsApp, e.g., the need to change phone numbers regularly. We finally experiment with on-device classification to automate the detection of junk, whilst respecting end-to-end encryption.

Since the inception of the first web page three decades back, the Web has evolved considerably, from static HTML pages in the beginning to the dynamic web pages of today, from mainly the text-based pages of the 1990s to today’s multimedia rich pages, etc.. Although much of this is known anecdotally, to our knowledge, there is no quantitative documentation of the extent and timing of these changes. This paper attempts to address this gap in the literature by looking at the top 100 Alexa websites for over 25 years from the Internet Archive or the “Wayback Machine”, archive.org. We study the changes in popularity, from Geocities and Yahoo! in the mid-to-late 1990s to the likes of Google, Facebook, and Tiktok of today. We also look at different categories of websites and their popularity over the years and find evidence for the decline in popularity of news and education-related websites, which have been replaced by streaming media and social networking sites. We explore the emergence and relative prevalence of different MIME-types (text vs. image vs. video vs. javascript and json) and study whether the use of text on the Internet is declining.

Connected Cars (CCs) and vehicle-to-everything (V2X) use cases require stringent reliability for safety and non-safety uses. With increasing network softwarisation, it has become easier to use multiple, redundant connectivity options instead of relying on a single network connectivity. But where should these redundant connections be managed? Is it at a network provider’s core network - i.e., supply side managed (SSM) - or at the CC - i.e., demand side managed (DSM)? In our work, we investigate the use of SSM and DSM for CCs on four separate days and across 800 kilometers of major / minor roads in South East England. For Day 1, we captured performance indicators, and determined hypothetical multi-operator configurations for four U.K. providers and a global Universal SIM. For Day 2, 3 & 4, we built and deployed a test-bed to actually implement network switching and understand performance (incl. for TCP & UDP) either on the road or in a stationary location. Based on our results, we make three contributions. First, we show that DSM can deliver superior performance for CCs more than any individual network (up to 28 percentage points in a hypothetical scenario), or SSM which had up to 4.8x longer page load times. Second, unlike other smartphone-only studies, our system-level study demonstrates that improvements of at least 12% can be obtained in a practical DSM field implementation for a CC. Third, we confirm that the advantage of DSM in a field implementation is higher for UDP traffic (23% better latency) compared to TCP (13%).

Decentralising the Web is a desirable but challenging goal. One particular challenge is achieving decentralised content moderation in the face of various adversaries (e.g. trolls). To overcome this challenge, many Decentralised Web (DW) implementations rely on federation policies. Administrators use these policies to create rules that ban or modify content that matches specific rules. This, however, can have unintended consequences for many users. In this paper, we present the first study of federation policies on the DW, their in-the-wild usage, and their impact on users. We identify how these policies may negatively impact "innocent" users and outline possible solutions to avoid this problem in the future.

Ella Guest, Bertie Vidgen, Alexandros Mittos, Nishanth Sastry, Gareth Tyson, Helen Margetts. Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume. 2021.

The viability of new mission-critical networked applications such as connected cars or remote surgery is heavily dependent on the availability of truly customized network services at a Quality of Service (QoS) level that both the network operator and the customer can agree on. This is difficult to achieve in today’s mainly "best effort" Internet. Even if a level of service were to be agreed upon between a consumer and an operator, it is important for both parties to be able to scalably and impartially monitor the quality of service delivered in order to enforce the service level agreement (SLA). Building upon a recently proposed architecture for automated negotiation of SLAs using smart contracts, we develop a low overhead solution for monitoring these SLAs and arranging automated payments based on the smart contracts. Our solution uses cryptographically secure bloom filters to create succinct summaries of the data exchanged over fine-grained epochs. We then use a state channel-based design for both parties to quickly and scalably agree and sign off on the data that was delivered in each epoch, making it possible to monitor and enforce at run time the agreed upon QoS levels.

As adoption of connected cars (CCs) grows, the expectation is that 5G will better support safety-critical vehicle-to-everything (V2X) use cases. Operationally, most relationships between cellular network providers and car manufacturers or users are exclusive, providing a single network connectivity, with at best an occasional option of a back-up plan if the single network is unavailable. We question if this setup can provide QoS assurance for V2X use cases. Accordingly, in this paper, we investigate the role of redundancy in providing QoS assurance for cellular connectivity for CCs. Using our bespoke Android measurement app, we did a drive-through test on 380 kilometers of major and minor roads in South East England. We measured round trip times, jitter, page load times, packet loss, network type, uplink speed and downlink speeds on the four UK networks for 14 UK-centric websites every five minutes. In addition, we did the same measurement using a much more expensive universal SIM card provider that promises to fall back on any of the four UK networks to assure reliability. By comparing actual performance on the best performing network versus the universal SIM, and then projected performance of a two/three/four multi-operator setup, we make three major contributions. First, the use of redundant multi-connectivity, especially if managed by the demand-side, can deliver superior performance (up to 28 percentage points in some cases). Second, despite costing 95x more per GB of data, the universal SIM performed worse than the best performing network except for uplink speed, highlighting how the choice of parameter to monitor can influence operational decisions. Third, any assessment of CC connectivity reliability based on availability is sub-optimal as it can hide significant under-performance.

IoT-driven smart societies are modern service-networked ecosystems, whose proper functioning is hugely based on the success of supply chain relationships. Robust security is still a big challenge in such ecosystems, catalyzed primarily by naive cyber-security practices (e.g., setting default IoT device passwords) on behalf of the ecosystem managers, i.e., users and organizations. This has recently led to some catastrophic malware-driven DDoS and ransomware attacks (e.g., the Mirai and WannaCry attacks). Consequently, markets for commercial third-party cyber-risk management (CRM) services (e.g., cyber-insurance) are steadily but sluggishly gaining traction with the rapid increase of IoT deployment in society, and provides a channel for ecosystem managers to transfer residual cyber-risk post attack events. Current empirical studies have shown that such residual cyber-risks affecting smart societies are often heavy-tailed in nature and exhibit tail dependencies. This is both, a major concern for a profit-minded CRM firm that might normally need to cover multiple such dependent cyber-risks from different sectors (e.g., manufacturing and energy) in a service-networked ecosystem, and a good intuition behind the sluggish market growth of CRM products. In this article, we provide: 1) a rigorous general theory to elicit conditions on (tail-dependent) heavy-tailed cyber-risk distributions under which a risk management firm might find it (non)sustainable to provide aggregate cyber-risk coverage services for smart societies and 2) a real-data-driven numerical study to validate claims made in theory assuming boundedly rational cyber-risk managers, alongside providing ideas to boost markets that aggregate dependent cyber-risks with heavy-tails. To the best of our knowledge, this is the only complete general theory till date on the feasibility of aggregate CRM.

As authors of our recently accepted article:Aggregate Cyber-Risk Management in the IoT Age: Cautionary Statistics for (Re)Insurers and Likes, published in the IEEE IoT Journal, we regret that we have found a few errors in the numerical evaluation setup of the works in[1]and[2]that we had borrowed for our accepted paper. In this correction statement, we describe the errors in detail, correct it, and present our revised results with a renewed experimental setup, hoping it to replace the existing incorrect numerical results in the accepted paper. We apologize for the inconvenience caused to the reader. We emphasize that the numerical evaluation section does not in any way hamper the theoretical contributions in this article, and was initially only meant to provide some empirical evidence for whether the theory proposed in this article generalizes to behavioral settings introduced in[2].

Video accounts for a large proportion of traffic on the Internet. Understanding its geographical viewing patterns is extremely valuable for the design of Internet ecosystems for content delivery, recommendation and ads. While previous works have addressed this problem at coarse-grain scales (e.g., national), the urban-scale geographical patterns of video access have never been revealed. To this end, this article aims to investigate the problem that whether there exists distinct viewing patterns among the neighborhoods of a large-scale city. To achieve this, we need to address several challenges including unknown of patterns profiles, complicate urban neighborhoods, and comprehensive viewing features. The contributions of this article include two aspects. First, we design a framework to automatically identify geographical video viewing patterns in urban neighborhoods. Second, by using a dataset of two months real video requests in Shanghai collected from one major ISP of China, we make a rigorous analysis of video viewing patterns in Shanghai. Our study reveals the following important observations. First, there exists four prevalent and distinct patterns of video access behavior in urban neighborhoods, which are corresponding to four different geographical contexts: downtown residential, office, suburb residential and hybrid regions. Second, there exists significant features that distinguish different patterns, e.g., the probabilities of viewing TV plays at midnight, and viewing cartoons at weekends can distinguish the two viewing patterns corresponding to downtown and suburb regions.

In-app advertising is a multi-billion dollar industry that is an essential part of the current digital ecosystem, and is amenable to sensitive consumer information often being sold downstream without the knowledge of consumers, and in many cases to their annoyance. While this practice, in cases, may result in long-term benefits for the consumers, it can result in serious information privacy (IP) breaches of very significant impact (e.g., breach of genetic data) in the short term. The question we raise through this article is: does the type of information being traded downstream play a role in the degree of IP risks generated? We investigate two general (one-many) information trading market structures between a single data aggregating seller (e.g., enterprise app) and multiple competing buyers (e.g., ad-networks, retailers), distinguished by mutually exclusive and privacy sanitized aggregated consumer data (information) types: (i) data entailing strategically complementary actions among buyers and (ii) data entailing strategically substituting actions among buyers. Our primary question of interest here is: trading which type of data might pose less information privacy risks for society? To this end, we show that at market equilibrium IP trading markets exhibiting strategic substitutes between buying firms pose lesser risks for IP in society, primarily because the `substitutes' setting, in contrast to the `complements' setting, economically incentivizes appropriate consumer data distortion by the seller in addition to restricting the proportion of buyers to which it sells. Moreover, we also show that irrespective of the data type traded by the seller, the likelihood of improved IP in society is higher if there is purposeful or free-riding based transfer/leakage of data between buying firms. This is because the seller finds itself economically incentivized to restrict the release of sanitized consumer data with respect to the span of its buyer space, as well as in improved data quality.

In online debates, as in offline ones, individual utterances or arguments support or attack each other, leading to some subset of arguments (potentially from different sides of the debate) being considered more relevant than others. However, online conversations are much larger in scale than offline ones, with often hundreds of thousands of users weighing in, collaboratively forming large trees of comments by starting from an original post and replying to each other. In large discussions, readers are often forced to sample a subset of the arguments being put forth. Since such sampling is rarely done in a principled manner, users may not read all the relevant arguments to get a full picture of the debate from a sample. This article is interested in answering the question of how users should sample online conversations to selectively favour the currently justified or accepted positions in the debate. We apply techniques from argumentation theory and complex networks to build a model that predicts the probabilities of the normatively justified arguments given their location in idealised online discussions of comments and replies, which we represent as trees. Our model shows that the proportion of replies that are supportive, the distribution of the number of replies that comments receive, and the locations of comments that do not receive replies (i.e., the “leaves” of the reply tree) all determine the probability that a comment is a justified argument given its location. We show that when the distribution of the number of replies is homogeneous along the tree length, for acrimonious discussions (with more attacking comments than supportive ones), the distribution of justified arguments depends on the parity of the tree level, which is the distance from the root expressed as number of edges. In supportive discussions, which have more supportive comments than attacks, the probability of having justified comments increases as one moves away from the root. For discussion trees that have a non-homogeneous in-degree distribution, for supportive discussions we observe the same behaviour as before, while for acrimonious discussions we cannot observe the same parity-based distribution. This is verified with data obtained from the online debating platform Kialo. By predicting the locations of the justified arguments in reply trees, we can therefore suggest which arguments readers should sample, to grasp the currently accepted opinions in such discussions. Our models have important implications for the design of future online debating platforms.

This paper aims to understand how third-party ecosystems have developed in four different countries: UK, China, AU, US. We are interested in how wide a view a given third-party player may have, of an individual user's browsing history over a period of time, and of the collective browsing histories of a cohort of users in each of these countries. We study this by utilizing two complementary approaches: the first uses lists of the most popular websites per country, as determined by Alexa.com. The second approach is based on the real browsing histories of a cohort of users in these countries. Our larger continuous user data collection spans over a year. Some universal patterns are seen, such as more third parties on more popular websites, and a specialization among trackers, with some trackers present in some categories of websites but not others. However, our study reveals several unexpected country-specific patterns: China has a home-grown ecosystem of third-party operators in contrast with the UK, whose trackers are dominated by players hosted in the US. UK trackers are more location sensitive than Chinese trackers. One important consequence of these is that users in China are tracked lesser than users in the UK. Our unique access to the browsing patterns of a panel of users provides a realistic insight into third party exposure, and suggests that studies which rely solely on Alexa top ranked websites may be over estimating the power of third parties, since real users also access several niche interest sites with lesser numbers of many kinds of third parties, especially advertisers.

The nature of information gathering and dissemination has changed dramatically over the past 20 years as traditional media sources are increasingly being replaced by a cacaphony of social media channels. Despite this, society still expects to disseminate its critical information via traditional news sources. Public Warning Systems (PWS) exist, but concerns about spamming users with irrelevant warnings mean that mostly only life threatening emergency warnings are delivered via PWS. We argue that it is time for society to upgrade its infrastructure for critical information services (CIS) and that a smartphone app system can provide a standardised, less-intrusive user interface to deliver CIS, especially if the traffic for the app is prioritised during congestion periods. Accordingly, we make three contributions in this paper. Firstly, using network parameters from our longitudinal measurements of network performance in Central London (an area of high user traffic), we show, with simulations, that reserving some bandwidth exclusively for CIS could assure QoS for CIS without significant degradation for other services. Secondly, we provide a conceptual design of a 999 CIS app, which can mimic the current 999 voice system and can be built using 3GPP defined systems. Thirdly, we identify the stakeholder relationships with industry partners and policymakers that can help to deliver a CIS system that is fit for purpose for an increasingly smartphone-based society.

Websites with hyper-partisan, left or right-leaning focus offer content that is typically biased towards the expectations of their target audience. Such content often polarizes users, who are repeatedly primed to specific (extreme) content, usually reflecting hard party lines on political and socio-economic topics. Though this polarization has been extensively studied with respect to content, it is still unknown how it associates with the online tracking experienced by browsing users, especially when they exhibit certain demographic characteristics. For example, it is unclear how such websites enable the ad-ecosystem to track users based on their gender or age.

The exploding volumes of mobile video traffic call for deploying content caches inside mobile operator networks. With in-network caching, users' requests for popular content can be served from a content cache deployed at mobile gateways in vicinity to the end user. This inherently reduces the load on the content servers and the backbone of operator's network. In light of the increasing trend in virtualization of network functions, we propose a cost-effective caching as a service (CaaS) framework for virtual video caching in 5G mobile networks. In order to evaluate the pros and cons of our CaaS approach, we formulate two virtual caching problems, namely maximum return on investment(MRI) and maximum offloaded traffic (MOT). MRI aims at maximizing return on caching investment by finding the best trade-off between the cost of cache storage and bandwidth savings from caching video contents in the mobile network operator (MNO)'s cloud. Likewise, MOT aims to maximize the traffic offloaded from the MNO's core and backhaul within given budget constraints. More specifically, taking the popularity and size of video contents into account, MRI and MOT aim to find the optimal caching tables which maximize the ratio of transmission bandwidth cost to storage cost and the offloaded traffic for a given budget, respectively. We reduce the complexity of the proposed problem formulated as a binary-integer programming (BIP) by using canonical duality theory (CDT). Experimental results obtained using the invasive weed optimization (IWO) have shown significant performance enhancement of the proposed system in terms of return on investment, quality, offloaded traffic, and storage efficiency.

On-demand video accounts for the majority of wireless data traffic. Video distribution schemes based on caching combined with device-to-device (D2D) communications promise order-of-magnitude greater spectral efficiency for video delivery, but hinge on the principle of concentrated demand distributions. This paper presents, for the first time, the analysis and evaluations of the throughput-outage tradeoff of such schemes based on measured cellular demand distributions. In particular, we use a dataset with more than 100 million requests from the BBC iPlayer, a popular video streaming service in the U.K., as the foundation of the analysis and evaluations. We present an achievable scaling law based on the practical popularity distribution, and show that such scaling law is identical to those reported in the literature. We find that also for the numerical evaluations based on a realistic setup, order-of-magnitude improvements can be achieved. Our results indicate that the benefits promised by the caching-based D2D in the literature could be retained for cellular networks in practice.

The Decentralised Web (DW) has recently seen a renewed momentum, with a number of DW platforms like Mastodon, PeerTube, and Hubzilla gaining increasing traction. These offer alternatives to traditional social networks like Twitter, YouTube, and Facebook, by enabling the operation of web infrastructure and services without centralised ownership or control. Although their services differ greatly, modern DW platforms mostly rely on two key innovations: first, their open source software allows anybody to setup independent servers ("instances") that people can sign-up to and use within a local community; and second, they build on top of federation protocols so that instances can mesh together, in a peer-to-peer fashion, to offer a globally integrated platform.

Caching of video files at the wireless edge, i.e., at the base stations or on user devices, is a key method for improving wireless video delivery. While global popularity distributions of video content have been investigated in the past and used in a variety of caching algorithms, this paper investigates the statistical modeling of the individual user preferences. With individual preferences being represented by probabilities, we identify their critical features and parameters and propose a novel modeling framework by using a genre-based hierarchical structure as well as a parameterization of the framework based on an extensive real-world data set. Besides, the correlation analysis between parameters and critical statistics of the framework is conducted. With the framework, an implementation recipe for generating practical individual preference probabilities is proposed. By comparing with the underlying real data, we show that the proposed models and generation approach can effectively characterize the individual preferences of users for video content.

P2P sharing amongst consumers has been proposed as a way to decrease load on Content Delivery Networks. This paper develops an analytical model that shows an additional benefit of sharing content locally: Selecting close by peers to share content from leads to shorter paths compared to traditional CDNs, decreasing the overall carbon footprint of the system. Using data from a month-long trace of over 3 million monthly users in London accessing TV shows online, we show that local sharing can result in a decrease of 24-48% in the system-wide carbon footprint of online video streaming, despite various obstacle factors that can restrict swarm sizes. We confirm the robustness of the savings by using realistic energy parameters drawn from two widely used settings. We also show that if the energy savings of the CDN servers are transferred as carbon credits to the end users, over 70% of users can become carbon positive, i.e., are able to support their content consumption without incurring any carbon footprint, and are able to offset their other carbon consumption. We suggest carbon credit transfers from CDNs to end users as a novel way to incentivise participation in peer-assisted content delivery.

The era of live-broadcast is back but with two major changes. First, unlike traditional TV broadcasts, content is now streamed over the Internet enabling it to reach a wider audience. Second, due to various user-generated content platforms it has become possible for anyone to get involved, streaming their own content to the world. This emerging trend of going live usually happens via social platforms, where users perform live social broadcasts predominantly from their mobile devices, allowing their friends (and the general public) to engage with the stream in real-time. With the growing popularity of such platforms, the burden on the current Internet infrastructure is therefore expected to multiply. With this in mind, we explore one such prominent platform - Facebook Live. We gather 3TB of data, representing one month of global activity and explore the characteristics of live social broadcast. From this, we derive simple yet effective principles which can decrease the network burden. We then dissect global and hyper-local properties of the video while on-air, by capturing the geography of the broadcasters or the users who produce the video and the viewers or the users who interact with it. Finally, we study the social engagement while the video is live and distinguish the key aspects when the same video goes on-demand. A common theme throughout the paper is that, despite its name, many attributes of Facebook Live deviate from both the concepts of live and broadcast.

Many adult content websites incorporate social networking features. Although these are popular, they raise significant challenges, including the potential for users to "catfish", i.e., to create fake profiles to deceive other users. This paper takes an initial step towards automated catfish detection. We explore the characteristics of the different age and gender groups, identifying a number of distinctions. Through this, we train models based on user profiles and comments, via the ground truth of specially verified profiles. When applying our models for age and gender estimation to unverified profiles, 38% of profiles are classified as lying about their age, and 25% are predicted to be lying about their gender. The results suggest that women have a greater propensity to catfish than men. Our preliminary work has notable implications on operators of such online social networks, as well as users who may worry about interacting with catfishes.

Caching of video files at the wireless edge, i.e., at the base stations or on user devices, is a key method for improving wireless video delivery. While global popularity distributions of video content have been investigated in the past, and used in a variety of caching algorithms, this paper investigates the statistical modeling of the individual user preferences. With individual preferences being represented by probabilities, we identify their critical features and parameters and propose a novel modeling framework as well as a parameterization of the framework based on an extensive real-world data set. Besides, an implementation recipe for generating practical individual preference probabilities is proposed. By comparing with the underlying real data, we show that the proposed models and generation approach can effectively characterize individual preferences of users for video content.

Vehicular CrowdSensing (VCS) is an emerging solution designed to remotely collect data from smart vehicles. It enables a dynamic and large-scale phenomena monitoring just by exploring the variety of technologies which have been embedded in modern cars. However, VCS applications might generate a huge amount of data traffic between vehicles and the remote monitoring center, which tends to overload the LTE networks. In this paper, we describe and analyze a gEo-clUstering approaCh for Lte vehIcular crowDsEnsing dAta offloadiNg (EUCLIDEAN). It takes advantage of opportunistic vehicle-to-vehicle (V2V) communications to support the VCS data upload process, preserving, as much as possible, the cellular network resources. In general, it is shown from the presented results that our proposal is a feasible and an effective scheme to reduce up to 92.98% of the global demand for LTE transmissions while performing vehicle-based sensing tasks in urban areas. The most encouraging results were perceived mainly under high-density conditions (i.e., above 125 vehicles/km2), where our solution provides the best benefits in terms of cellular network data offloading.

Peer-assisted content delivery networks have recently emerged as an economically viable alternative to traditional content delivery approaches: the feasibility studies conducted for several large content providers suggested a remarkable potential of peer-assisted content delivery networks to reduce the burden of user requests on content delivery servers and several commercial peer-assisted deployments have been recently introduced. Yet there are many technical and commercial challenges which question the future of peer-assisted solutions in industrial settings. This includes among others unreliability of peer-to-peer networks, the lack of incentives for peers’ participation, and copyright issues. In this paper, we carefully review and systematize this ongoing debate around the future of peer-assisted networks and propose a novel taxonomy to characterize the research and industrial efforts in the area. To this end, we conduct a comprehensive survey of the last decade in the peer-assisted content delivery research and devise a novel taxonomy to characterize the identified challenges and the respective proposed solutions in the literature. Our survey includes a thorough review of the three very large scale feasibility studies conducted for BBC iPlayer, MSN Video and Conviva, five large commercial peer-assisted CDNs - Kankan, LiveSky, Akamai NetSession, Spotify, Tudou - and a vast scope of technical papers. We focus both on technical challenges in deploying peer-assisted solutions and also on non-technical challenges caused due to heterogeneity in user access patterns and distribution of resources among users as well as commercial feasibility related challenges attributed to the necessity of accounting for the interests and incentives of Internet Service Providers, End-Users and Content Providers. The results of our study suggest that many of technical challenges for implementing peer-assisted content delivery networks on an industrial scale have been already addressed in the literature, whereas a problem of finding economically viable solutions to incentivize participation in peer-assisted schemes remains an open issue to a large extent. Furthermore, the emerging Internet of Things (IoT) is expected to enable expansion of conventional CDNs to a broader network of connected devices through machine to machine communication.

The exploding volumes of mobile video traffic call for deploying content caches inside mobile operators network. With in-network caching, users' requests for popular content can be served from a content cache deployed at mobile gateways in vicinity to the end user, therefore considerably reducing the load on the content servers and the backbone of operator's network. In practice, content caches can be installed at multiple levels inside an operator's network (e.g., serving gateway, packet data network gateway, RAN, etc.), leading to an idea of hierarchical in-network video caching. In order to evaluate the pros and cons of hierarchical caching, in this paper we formulate a cache provisioning problem which aims to find the best tradeoff between the cost of cache storage and bandwidth savings from hierarchical caching. More specifically, we aim to find the optimal size of video caches at different layers of a hierarchical in-network caching architecture which minimizes the ratio of transmission bandwidth cost to storage cost. We overcome the complexity of our problem which is formulated as a binary-integer programming (BIP) by using canonical duality theory (CDT). Numerical results obtained using the invasive weed optimization (IWO) show that important gains can be achieved, with benefit-cost ratio and cost efficiency improvements of more than 43% and 38%, respectively.

TV White Spaces (TVWS) and associated spectrum sharing mechanisms represent key means of realizing necessary prime-frequency spectrum for future wireless communication systems. We have been leading a major trial of TVWS technology within the Ofcom TV White Spaces Pilot. As one aspect of the work of our trial, we have investigated solutions for aggregation in TVWS and as part of that the performance of InterDigital White Space Devices (WSDs), capable of aggregating a IEEE 802.11 enabled technology for operation in up to 4 TVWS channels, non-contiguously as well as contiguously. This paper reports on some of our assessment of aggregation in TVWS, as well as our assessment of the InterDigital WSDs. It reports on the white space channel availabilities that can be achieved through aggregation, based on a real implementation of a WSD exhaustively testing a large area of England with a high resolution. The considerable benefit that is achieved through allowing non-contiguous aggregation as compared with contiguous-only aggregation is shown. Further, this paper assesses the TCP and UDP throughput performances of the InterDigital WSDs against the number of channels aggregated and received signal powers, in highly controlled scenarios. Statistics on performance of the WSDs for the studied large area of England are derived based on this. These results are compared with theoretical similar WSDs with one major difference that they can only achieve contiguous channel aggregation. Results show almost a doubling of capacity through non-contiguous aggregation with the InterDigital WSDs; this performance benefit would be increased significantly if more than 4 channels were supported for aggregation.

On most current websites untrustworthy or spammy identities are easily created. Existing proposals to detect untrustworthy identities rely on reputation signals obtained by observing the activities of identities over time within a single site or domain; thus, there is a time lag before which websites cannot easily distinguish attackers and legitimate users. In this paper, we investigate the feasibility of leveraging information about identities that is aggregated across multiple domains to reason about their trustworthiness. Our key insight is that while honest users naturally maintain identities across multiple domains (where they have proven their trustworthiness and have acquired reputation over time), attackers are discouraged by the additional effort and costs to do the same. We propose a flexible framework to transfer trust between domains that can be implemented in today's systems without significant loss of privacy or significant implementation overheads.

Mobile data offloading can greatly decrease the load on and usage of current and future cellular data networks by exploiting opportunistic and frequent access to Wi-Fi connectivity. Unfortunately, Wi-Fi access from mobile devices can be difficult during typical work commutes, e.g., via trains or cars on highways. In this paper, we propose a new approach: to preload the mobile device with content that a user might be interested in, thereby avoiding the need for cellular data access. We demonstrate the feasibility of this approach by developing a supervised machine learning model that learns from user preferences for different types of content, and propensity to be guided by the user interface of the player, and predictively preload entire TV shows. Testing on a data set of nearly 3.9 million sessions from all over the U.K. to BBC TV shows, we find that predictive preloading can save over 71% of the mobile data for an average user.

Today, the Internet is a large multimedia delivery infrastructure, with websites such as YouTube appearing at the top of most measurement studies. However, most traffic studies have ignored an important domain: adult multimedia distribution. Whereas, traditionally, such services were provided primarily via bespoke websites, recently these have converged towards what is known as “Porn 2.0”. These services allow users to upload, view, rate, and comment on videos for free (much like YouTube). Despite their scale, we still lack even a basic understanding of their operation. This article addresses this gap by performing a large-scale study of one of the most popular Porn 2.0 websites: YouPorn. Our measurements reveal a global delivery infrastructure that we have repeatedly crawled to collect statistics (on 183k videos). We use this data to characterise the corpus, as well as to inspect popularity trends and how they relate to other features, for example, categories and ratings. To explore our discoveries further, we use a small-scale user study, highlighting key system implications.

The last 5 years have seen a dramatic shift in media distribution. For decades, TV and radio were solely provisioned using push-based broadcast technologies, forcing people to adhere to fixed schedules. The introduction of catch-up services, however, has now augmented such delivery with online pull-based alternatives. Typically, these allow users to fetch content for a limited period after initial broadcast, allowing users flexibility in accessing content. Whereas previous work has investigated both of these technologies, this paper explores and contrasts them, focusing on the network consequences of moving towards this multifaceted delivery model. Using traces from nearly 6 million users of BBC iPlayer, one of the largest catch-up TV services, we study this shift from push- to pull-based access. We propose a novel technique for unifying both push- and pull-based delivery: the Speculative Content Offloading and Recording Engine (SCORE). SCORE operates as a set-top box, which interacts with both broadcast push and online pull services. Whenever users wish to access media, it automatically switches between these distribution mechanisms in an attempt to optimize energy efficiency and network resource utilization. SCORE also can predict user viewing patterns, automatically recording certain shows from the broadcast interface. Evaluations using our BBC iPlayer traces show that, based on parameter settings, an oracle with complete knowledge of user consumption can save nearly 77% of the energy, and over 90% of the peak bandwidth, of pure IP streaming. Optimizing for energy consumption, SCORE can recover nearly half of both traffic and energy savings.

In search of scalable solutions, CDNs are exploring P2P support. However, the benefits of peer assistance can be limited by various obstacle factors such as ISP friendliness - requiring peers to be within the same ISP, bitrate stratification - the need to match peers with others needing similar bitrate, and partial participation - some peers choosing not to redistribute content. This work relates potential gains from peer assistance to the average number of users in a swarm, its capacity, and empirically studies the effects of these obstacle factors at scale, using a month-long trace of over 2 million users in London accessing BBC shows online. Results indicate that even when P2P swarms are localised within ISPs, up to 88% of traffic can be saved. Surprisingly, bitrate stratification results in 2 large sub-swarms and does not significantly affect savings. However, partial participation, and the need for a minimum swarm size do affect gains. We investigate improvements to gain from increasing content availability through two well-studied techniques: content bundling-combining multiple items to increase availability, and historical caching of previously watched items. Bundling proves ineffective as increased server traffic from larger bundles outweighs benefits of availability, but simple caching can considerably boost traffic gains from peer assistance.

Using nine months of access logs comprising 1.9 Billion sessions to BBC iPlayer, we survey the UK ISP ecosystem to understand the factors affecting adoption and usage of a high bandwidth TV streaming application across different providers. We find evidence that connection speeds are important and that external events can have a huge impact for live TV usage. Then, through a temporal analysis of the access logs, we demonstrate that data usage caps imposed by mobile ISPs significantly affect usage patterns, and look for solutions. We show that product bundle discounts with a related fixed-line ISP, a strategy already employed by some mobile providers, can better support user needs and capture a bigger share of accesses. We observe that users regularly split their sessions between mobile and fixed-line connections, suggesting a straightforward strategy for offloading by speculatively pre-fetching content from a fixed-line ISP before access on mobile devices.

TV White Spaces (TVWS) technology allows wireless devices to opportunistically use locally-available TV channels enabled by a geolocation database. The UK regulator Ofcom has initiated a pilot of TVWS technology in the UK. This paper concerns a large- scale series of trials under that pilot. The purposes are to test aspects of white space technology, including the white space device and geolocation database interactions, the validity of the channel availability/powers calculations by the database and associated interference effects on primary services, and the performances of the white space devices, among others. An additional key purpose is to perform research investigations such as on aggregation of TVWS resources with conventional resources and also aggregation solely within TVWS, secondary coexistence issues and means to mitigate such issues, and primary coexistence issues under challenging deployment geometries, among others. This paper provides an update on the trials, giving an overview of their objectives and characteristics, some aspects that have been covered, and some early results and observations.

Everyday, millions of users save content items for future use on sites like Pinterest, by ''pinning'' them onto carefully categorised personal pinboards, thereby creating personal taxonomies of the Web. This paper seeks to understand Pinterest as a distributed human computation that categorises images from around the Web. We show that despite being categorised onto personal pinboards by individual actions, there is a generally a global agreement in implicitly assigning images into a coarse-grained global taxonomy of 32 categories, and furthermore, users tend to specialise in a handful of categories. By exploiting these characteristics, and augmenting with image-related features drawn from a state-of-the-art deep convolutional neural network, we develop a cascade of predictors that together automate a large fraction of Pinterest actions. Our end-to-end model is able to both predict whether a user will repin an image onto her own pinboard, and also which pinboard she might choose, with an accuracy of 0.69 ([email protected] of 0.75).

On-demand video streaming dominates today's Internet traffic mix. For instance, Netflix constitutes a third of the peak time traffic in the USA. Nearly half of UK online households have accessed BBC's shows through its on-demand streaming interface, BBC iPlayer. Using UK-wide traces from BBC iPlayer as a case study, this talk will characterise users' content consumption at scale and discuss techniques that can be deployed at the edge by users to substantially decrease the load on the Internet. We will survey both well-known techniques such as peer-assisted VoD, studying whether it works at scale, as well as new edge-caching mechanisms that can potentially be deployed today. We will conclude by exploring new directions for content-centric network architectures, to address the roots of the pain points observed in our user workload, in a “clean” fashion.

How does one develop a new online community that is highly engaging to each user and promotes social interaction? A number of websites offer friend-finding features that help users bootstrap social networks on the website by copying links from an established network like Facebook or Twitter. This paper quantifies the extent to which such social bootstrapping is effective in enhancing a social experience of the website. First, we develop a stylised analytical model that suggests that copying tends to produce a giant connected component (i.e., a connected community) quickly and preserves properties such as reciprocity and clustering, up to a linear multiplicative factor. Second, we use data from two websites, Pinterest and Last.fm, to empirically compare the subgraph of links copied from Facebook to links created natively. We find that the copied subgraph has a giant component, higher reciprocity and clustering, and confirm that the copied connections see higher social interactions. However, the need for copying diminishes as users become more active and influential. Such users tend to create links natively on the website, to users who are more similar to them than their Facebook friends. Our findings give new insights into understanding how bootstrapping from established social networks can help engage new users by enhancing social interactivity.

The Internet has evolved into a huge video delivery infrastructure, with websites such as YouTube and Netflix appearing at the top of most traffic measurement studies. However, most traffic studies have largely kept silent about an area of the Internet that (even today) is poorly understood: adult media distribution. Whereas ten years ago, such services were provided primarily via peer-to-peer file sharing and bespoke websites, recently these have converged towards what is known as ``Porn 2.0''. These popular web portals allow users to upload, view, rate and comment videos for free. Despite this, we still lack even a basic understanding of how users interact with these services. This paper seeks to address this gap by performing the first large-scale measurement study of one of the most popular Porn 2.0 websites: YouPorn. We have repeatedly crawled the website to collect statistics about 183k videos, witnessing over 60 billion views. Through this, we offer the first characterisation of this type of corpus, highlighting the nature of YouPorn's repository. We also inspect the popularity of objects and how they relate to other features such as the categories to which they belong. We find evidence for a high level of flexibility in the interests of its user base, manifested in the extremely rapid decay of content popularity over time, as well as high susceptibility to browsing order. Using a small-scale user study, we validate some of our findings and explore the infrastructure design and management implications of our observations.

"Catch-up", or on-demand access of previously broadcast TV content over the public Internet, constitutes a significant fraction of peak time network traffic. This paper analyses consumption patterns of nearly 6 million users of a nationwide deployment of a catch-up TV service, to understand the network support required. We find that catch-up has certain natural scaling properties compared to traditional TV: The on-demand nature spreads load over time, and users have much higher completion rates for content streams than previously reported. Users exhibit strong preferences for serialised content, and for specific genres.

In this paper we introduce LENS, a novel spam protection system based on the recipient's social network, which allows correspondence within the social circle to directly pass to the mailbox and further mitigates spam beyond social circles. The key idea in LENS is to select legitimate and authentic users, called Gatekeepers (GKs), from outside the recipients social circle and within pre-defined social distances. Unless a GK vouches for the emails of potential senders from outside the social circle of a particular recipient, those e-mails are prevented from transmission. In this way LENS drastically reduces the consumption of Internet bandwidth by spam. Using extensive evaluations, we show that LENS provides each recipient reliable email delivery from a large fraction of the social network. We also evaluate the computational complexity of email processing with LENS deployed on two Mail Servers (MSs) and compared it with the most popular content-based filter i.e SpamAssassin. LENS proved to be fast in processing emails (around 2-3 orders of magnitude better than SpamAssassin) and scales efficiently with increasing community size and GKs.

Pocket Switched Networks take advantage of social contacts to opportunistically create data paths over time. This work employs empirical traces to examine the effect of the human contact process on data delivery in such networks. The contact occurrence distribution is found to be highly uneven: contacts between a few node pairs occur too frequently, leading to inadequate mixing in the network, while the majority of contacts occur rarely, but are essential for global connectivity. This distribution of contacts leads to a significant variation in the fraction of node pairs that can be connected over time windows of similar duration. Good time windows tend to have a large clique of nodes that can all reach each other. It is shown that the clustering coefficient of the contact graph over a time window is a good predictor of achievable connectivity. We then examine all successful paths found by flooding and show that though delivery times vary widely, randomly sampling a small number of paths between each source and destination is sufficient to yield a delivery time distribution close to that of flooding over all paths. This result suggests that the rate at which the network can deliver data is remarkably robust to path failures.

The recently proposed packet switched network paradigm takes advantage of human social contacts to opportunistically create data paths over time. Our goal is to examine the effect of the human contact process on data delivery. We find that the contact occurrence distribution is highly uneven: contacts between a few node-pairs occur too frequently, leading to inadequate mixing in the network, while the majority of contacts are rare, and essential for connectivity. This distribution of contacts leads to a significant variation in performance over short time windows. We discover that the formation of a large clique core during the window is correlated with the fraction of data delivered, as well as the speed of delivery. We then show that the clustering co-efficient of the contact graph over a time window is a good predictor of performance during the window. Taken together, our findings suggest new directions for designing forwarding algorithms in ad-hoc or delay-tolerant networking schemes using humans as data mules.

Disparate algorithms are being designed to decide certain basic questions in opportunistic networks. This position paper describes a nascent idea that aims to provide a single framework to answer such questions. Inspired by the concept of a generic knowledge plane, we propose to study whether the information embodied in folksonomies can be used to make network decisions in opportunistic networks.

This work presents a comprehensive theoretical framework for memoryless window-based congestion control protocols that are designed to converge to fairness and efficiency. We first derive a necessary and sufficient condition for stepwise convergence to fairness. Using this, we show how fair window increase/decrease policies can be constructed from suitable pairs of monotonically nondecreasing functions. We generalize this to smooth protocols that converge over each congestion epoch. The framework also includes a simple method for incorporating TCP-friendliness. Well-studied congestion control protocols such as TCP, GAIMD, and Binomial congestion control can be constructed using this method. Thus, we provide a common framework for the analysis of such window-based protocols. We also present two new congestion control protocols for streaming media-like applications as examples of protocol design in this framework: The first protocol, LOG, has the objective of reconciling the smoothness requirement of an application with the need for a fast dynamic response to congestion. The second protocol, SIGMOID, guarantees a minimum bandwidth for an application but behaves exactly like TCP for large windows.

We present a comprehensive theoretical framework for window-based congestion control protocols that are designed to converge to fairness and efficiency. We first derive a sufficient condition for convergence to fairness. Using this, we show how fair window increase/decrease policies can be constructed from suitable pairs of monotonically non-decreasing functions. We show that well-studied protocols such as TCP, GAIMD (general additive-increase multiplicative-decrease) and binomial congestion control can be constructed using this method. Thus we provide a common framework for the analysis of such window-based protocols. To validate our approach, we present experimental results for a new TCP-friendly protocol, LOG, designed using this framework with the objective of reconciling the smoothness requirement of streaming media-like applications with the need for a fast dynamic response to congestion.