While social media offer great communication opportunities, they also increase the vulnerability of young people to threatening situations online. Recent studies report that cyberbullying constitutes a growing problem among youngsters. Successful prevention depends on the adequate detection of potentially harmful messages and the information overload on the Web requires intelligent systems to identify potential risks automatically. The focus of this paper is on automatic cyberbullying detection in social media text by modelling posts written by bullies, victims, and bystanders of online bullying. We describe the collection and fine-grained annotation of a cyberbullying corpus for English and Dutch and perform a series of binary classification experiments to determine the feasibility of automatic cyberbullying detection. We make use of linear support vector machines exploiting a rich feature set and investigate which information sources contribute the most for the task. Experiments on a hold-out test set reveal promising results for the detection of cyberbullying-related posts. After optimisation of the hyperparameters, the classifier yields an F1 score of 64% and 61% for English and Dutch respectively, and considerably outperforms baseline systems.

No abstract

The aims of this study were to examine, in a prospective, controlled way, the effect of the sperm deposition site in the oocyte and the mode of oolemma breakage in intracytoplasmic sperm injection (ICSI) on fertilization and embryo development rates. In the first trial (100 cycles in total), the spermatozoa were deposited further from the meiotic spindle (polar body at the 12 o'clock position) in half of the oocytes (n = 649), while in the other half (n = 605) the spermatozoa were deposited nearer to the meiotic spindle (polar body at the 6 o'clock position). In the second trial (6860 oocytes in 624 cycles), five different modes of membrane breakage (the reaction of the oolemma to the penetrating injection needle) at the moment of injection were noted: oolemma breakage, type A pricking once, no suction (n = 1401); type B, pricking once, small suction (n = 2761); type C, pricking once, long suction (n = 2310); type D, pricking twice or more, no or small amount of suction (n = 259); and type E, pricking twice or more, long suction (n = 129). No differences were observed between the 12 and 6 o'clock positions in the survival rate (90 and 90% respectively) and in the normal fertilization rates (78 and 77% respectively). Significantly more transfer quality embryos (< or = 50% fragmentation) were obtained in the 6 o'clock position group (83%) than in the 12 o'clock position group (79%). In the second trial, significantly lower survival rates were noted after membrane breakage type A (82%) than after breakages of types B, C, D and E (93, 92, 88 and 88% respectively). There were no significant differences present in the normal fertilization rates (70, 72, 70, 71 and 73% for types A-E respectively), but significantly more freeze quality embryos (< or = 20% fragmentation) were obtained after injection B (65%) than after injection types A, C, D and E (59, 61, 55 and 51% respectively). In conclusion, the site of sperm deposition in the oocyte does not influence the normal fertilization rate but does affect the embryo development rate. Furthermore, the mode of membrane breakage does not influence the normal fertilization rate but does affect oocyte survival and embryo development rates.

The detection of online cyberbullying has seen an increase in societal importance, popularity in research, and available open data. Nevertheless, while computational power and affordability of resources continue to increase, the access restrictions on high-quality data limit the applicability of state-of-the-art techniques. Consequently, much of the recent research uses small, heterogeneous datasets, without a thorough evaluation of applicability. In this paper, we further illustrate these issues, as we (i) evaluate many publicly available resources for this task and demonstrate difficulties with data collection. These predominantly yield small datasets that fail to capture the required complex social dynamics and impede direct comparison of progress. We (ii) conduct an extensive set of experiments that indicate a general lack of cross-domain generalization of classifiers trained on these sources, and openly provide this framework to replicate and extend our evaluation criteria. Finally, we (iii) present an effective crowdsourcing method: simulating real-life bullying scenarios in a lab setting generates plausible data that can be effectively used to enrich real data. This largely circumvents the restrictions on data that can be collected, and increases classifier performance. We believe these contributions can aid in improving the empirical practices of future research in the field.