Cyberbullying is a growing problem affecting more than half of all American teens. The main goal of this paper is to investigate fundamentally new approaches to understand and automatically detect and predict incidents of cyberbullying in Instagram, a media-based mobile social network. In this work, we have collected a sample data set consisting of Instagram images and their associated comments. We then designed a labeling study and employed human contributors at the crowd-sourced CrowdFlower website to label these media sessions for cyberbullying. A detailed analysis of the labeled data is then presented, including a study of relationships between cyberbullying and a host of features such as cyberaggression, profanity, social graph features, temporal commenting behavior, linguistic content, and image content. Using the labeled data, we further design and evaluate the performance of classifiers to automatically detect and predict incidents of cyberbullying and cyberaggression.

No abstract

High-throughput, low-cost, and accurate predictions of thermal properties of new materials would be beneficial in fields ranging from thermal barrier coatings and thermoelectrics to integrated circuits. To date, computational efforts for predicting lattice thermal conductivity (κ L ) have been hampered by the complexity associated with computing multiple phonon interactions. In this work, we develop and validate a semiempirical model for κ L by fitting density functional theory calculations to experimental data. Experimental values for κ L come from new measurements on SrIn 2 O 4 , Ba 2 SnO 4 , Cu 2 ZnSiTe 4 , MoTe 2 , Ba 3 In 2 O 6 , Cu 3 TaTe 4 , SnO, and InI as well as 55 compounds from across the published literature. To capture the anharmonicity in phonon interactions, we incorporate a structural parameter that allows the model to predict κ L within a factor of 1.5 of the experimental value across 4 orders of magnitude in κ L values and over a diverse chemical and structural phase space, with accuracy similar to or better than that of computationally more expensive models.

Decentralized and unstructured peer-to-peer networks such as Gnutella are attractive for certain applications because they require no centralized directories and no precise control over network topology or data placement. However, the flooding-based query algorithm used in Gnutella does not scale; each query generates a large amount of traffic and large systems quickly become overwhelmed by the queryinduced load. This paper explores, through simulation, various alternatives to Gnutella's query algorithm, data replication strategy, and network topology. We propose a query algorithm based on multiple random walks that resolves queries almost as quickly as Gnutella's flooding method while reducing the network traffic by two orders of magnitude in many cases. We also present simulation results on a distributed replication strategy proposed in [8]. Finally, we find that among the various network topologies we consider, uniform random graphs yield the best performance. jaded observers the explosive increase in Peer-to-Peer (P2P) network usage has been astounding. Within a few months of Napster's [16] introduction in 1999 the system had spread widely, and recent measurement data suggests that P2P applications are having a very significant and rapidly growing impact on Internet traffic [12,17]. It is important to study the performance and scalability of these P2P networks.Currently, there are several different architectures for P2P networks:Centralized: Napster and other similar systems have a constantly-updated directory hosted at central locations (e.g., the Napster web site). Nodes in the P2P network issue queries to the central directory server to find which other nodes hold the desired files. Such centralized approaches do not scale well and have single points of failure. Decentralized but Structured:These systems have no central directory server, but they have a significant amount of structure. By "structure" we mean that the P2P overlay topology (that is, the set of connections between P2P members) is tightly controlled and that files are placed not at random nodes but at specified locations that will make subsequent queries easier to satisfy. In "loosely structured" systems this placement of files is based on hints; the Freenet P2P network [11] is an example of such systems. In "highly structured" systems both the P2P network topology and the placement of files are precisely determined; this tightly controlled structure enables the system to satisfy queries very efficiently. There is a growing literature on highly structured P2P systems which support a hash-table-like interface; see [18,23,20,26]. Such highly structured P2P designs are quite prevalent in the research literature, but almost completely invisible on the current network. Moreover, it isn't clear how well such designs work with an extremely transient population of nodes, which seems to be a characteristic of the Napster community.Decentralized and Unstructured: These are systems in which there is neither a centralized directory nor any precise control...

Abstract. Advances in embedded systems and low-cost gas sensors are enabling a new wave of low-cost air quality monitoring tools. Our team has been engaged in the development of low-cost, wearable, air quality monitors (M-Pods) using the Arduino platform. These M-Pods house two types of sensors – commercially available metal oxide semiconductor (MOx) sensors used to measure CO, O3, NO2, and total VOCs, and NDIR sensors used to measure CO2. The MOx sensors are low in cost and show high sensitivity near ambient levels; however they display non-linear output signals and have cross-sensitivity effects. Thus, a quantification system was developed to convert the MOx sensor signals into concentrations. We conducted two types of validation studies – first, deployments at a regulatory monitoring station in Denver, Colorado, and second, a user study. In the two deployments (at the regulatory monitoring station), M-Pod concentrations were determined using collocation calibrations and laboratory calibration techniques. M-Pods were placed near regulatory monitors to derive calibration function coefficients using the regulatory monitors as the standard. The form of the calibration function was derived based on laboratory experiments. We discuss various techniques used to estimate measurement uncertainties. The deployments revealed that collocation calibrations provide more accurate concentration estimates than laboratory calibrations. During collocation calibrations, median standard errors ranged between 4.0–6.1 ppb for O3, 6.4–8.4 ppb for NO2, 0.28–0.44 ppm for CO, and 16.8 ppm for CO2. Median signal to noise (S / N) ratios for the M-Pod sensors were higher than the regulatory instruments: for NO2, 3.6 compared to 23.4; for O3, 1.4 compared to 1.6; for CO, 1.1 compared to 10.0; and for CO2, 42.2 compared to 300–500. By contrast, lab calibrations added bias and made it difficult to cover the necessary range of environmental conditions to obtain a good calibration. A separate user study was also conducted to assess uncertainty estimates and sensor variability. In this study, 9 M-Pods were calibrated via collocation multiple times over 4 weeks, and sensor drift was analyzed, with the result being a calibration function that included baseline drift. Three pairs of M-Pods were deployed, while users individually carried the other three. The user study suggested that inter-M-Pod variability between paired units was on the same order as calibration uncertainty; however, it is difficult to make conclusions about the actual personal exposure levels due to the level of user engagement. The user study provided real-world sensor drift data, showing limited CO drift (under −0.05 ppm day−1), and higher for O3 (−2.6 to 2.0 ppb day−1), NO2 (−1.56 to 0.51 ppb day−1), and CO2 (−4.2 to 3.1 ppm day−1). Overall, the user study confirmed the utility of the M-Pod as a low-cost tool to assess personal exposure.

Group recommendation, which makes recommendations to a group of users instead of individuals, has become increasingly important in both the workspace and people's social activities, such as brainstorming sessions for coworkers and social TV for family members or friends. Group recommendation is a challenging problem due to the dynamics of group memberships and diversity of group members. Previous work focused mainly on the content interests of group members and ignored the social characteristics within a group, resulting in suboptimal group recommendation performance.In this work, we propose a group recommendation method that utilizes both social and content interests of group members. We study the key characteristics of groups and propose (1) a group consensus function that captures the social, expertise, and interest dissimilarity among multiple group members; and (2) a generic framework that automatically analyzes group characteristics and constructs the corresponding group consensus function. Detailed user studies of diverse groups demonstrate the effectiveness of the proposed techniques, and the importance of incorporating both social and content interests in group recommender systems.

No abstract

Abstract. Advances in embedded systems and low-cost gas sensors are enabling a new wave of low cost air quality monitoring tools. Our team has been engaged in the development of low-cost wearable air quality monitors (M-Pods) using the Arduino platform. The M-Pods use commercially available metal oxide semiconductor (MOx) sensors to measure CO, O3, NO2, and total VOCs, and NDIR sensors to measure CO2. MOx sensors are low in cost and show high sensitivity near ambient levels; however they display non-linear output signals and have cross sensitivity effects. Thus, a quantification system was developed to convert the MOx sensor signals into concentrations. Two deployments were conducted at a regulatory monitoring station in Denver, Colorado. M-Pod concentrations were determined using laboratory calibration techniques and co-location calibrations, in which we place the M-Pods near regulatory monitors to then derive calibration function coefficients using the regulatory monitors as the standard. The form of the calibration function was derived based on laboratory experiments. We discuss various techniques used to estimate measurement uncertainties. A separate user study was also conducted to assess personal exposure and M-Pod reliability. In this study, 10 M-Pods were calibrated via co-location multiple times over 4 weeks and sensor drift was analyzed with the result being a calibration function that included drift. We found that co-location calibrations perform better than laboratory calibrations. Lab calibrations suffer from bias and difficulty in covering the necessary parameter space. During co-location calibrations, median standard errors ranged between 4.0–6.1 ppb for O3, 6.4–8.4 ppb for NO2, 0.28–0.44 ppm for CO, and 16.8 ppm for CO2. Median signal to noise (S/N) ratios for the M-Pod sensors were higher for M-Pods than the regulatory instruments: for NO2, 3.6 compared to 23.4; for O3, 1.4 compared to 1.6; for CO, 1.1 compared to 10.0; and for CO2, 42.2 compared to 300–500. The user study provided trends and location-specific information on pollutants, and affected change in user behavior. The study demonstrated the utility of the M-Pod as a tool to assess personal exposure.