Large-scale data sets of human behavior have the potential to fundamentally transform the way we fight diseases, design cities, or perform research. Metadata, however, contain sensitive information. Understanding the privacy of these data sets is key to their broad use and, ultimately, their impact. We study 3 months of credit card records for 1.1 million people and show that four spatiotemporal points are enough to uniquely reidentify 90% of individuals.We show that knowing the price of a transaction increases the risk of reidentification by 22%, on average. Finally, we show that even data sets that provide coarse information at any or all of the dimensions provide little anonymity and that women are more reidentifiable than men in credit card metadata.Large-scale data sets of human behavior have the potential to fundamentally transform the way we fight diseases, design cities, or perform research. Ubiquitous technologies create personal metadata on a very large scale. Our smartphones, browsers, cars, or credit cards generate infor-

Cyber Bullying, which often has a deeply negative impact on the victim, has grown as a serious issue among adolescents. To understand the phenomenon of cyber bullying, experts in social science have focused on personality, social relationships and psychological factors involving both the bully and the victim. Recently computer science researchers have also come up with automated methods to identify cyber bullying messages by identifying bullying-related keywords in cyber conversations. However, the accuracy of these textual feature based methods remains limited. In this work, we investigate whether analyzing social network features can improve the accuracy of cyber bullying detection. By analyzing the social network structure between users and deriving features such as number of friends, network embeddedness, and relationship centrality, we find that the detection of cyber bullying can be significantly improved by integrating the textual features with social network features.

No abstract

OBJECTIVES The purpose of this study was to use meta-analysis to establish which of the information available to the resident selection committee is associated with resident or doctor performance.METHODS Multiple electronic databases were searched to 4 September 2012. Two reviewers independently selected studies that met the present inclusion criteria and extracted data in duplicate; disagreement was resolved by consensus. Risk for bias was assessed using a customised bias assessment tool. Measures of association were converted to a common effect size (Hedges' g). Meta-analysis was performed using the random-effects model for each selection strategy and all outcomes without pooling. Sensitivity analysis for each selection strategy-outcome pair was performed with pooling of effect size.RESULTS Eighty studies involving a total of 41 704 participants were included in the metaanalysis. Seventeen different selection strategies and 17 outcomes were assessed across these studies. The strongest positive associations referred to examination-based selection strategies, such as the US Medical Licensing Examination (USMLE) Step 1, and examination-based outcomes, such as scores on intraining examinations. Moderate positive associations were present for medical school marks and both examination-based and subjective outcomes. Minimal or no associations were seen for the selection tools represented by interviews, reference letters and deans' letters.CONCLUSIONS Standardised examination performance and medical school grades show the strongest associations with current measures of doctor performance. Deans' letters, reference letters and interviews all show a lower than expected strength of association given the relative value often assigned to them during resident doctor selection. Objective selection strategies are potentially the most useful to residency selection committees based on current evaluative methods. However, reports in the literature of validated long-term doctor performance outcomes are scant.