Although it is under-studied relative to other social media platforms, YouTube arguably the largest and most engaging online consumption platform in world. Recently, YouTube's scale has fueled concerns that users are being radicalized via a combination of biased recommendations ostensibly apolitical "anti-woke" channels, both which have been claimed direct attention radical political content. Here we test this hypothesis using representative panel more than 300,000 Americans their...

Most real-world networks are incompletely observed. Algorithms that can accurately predict which links missing dramatically speedup the collection of network data and improve validity models. Many algorithms now exist for predicting links, given a partially observed network, but it has remained unknown whether single best predictor exists, how link predictability varies across methods from different domains, close to optimality current are. We answer these questions by systematically...

A common graph mining task is community detection, which seeks an unsupervised decomposition of a network into groups based on statistical regularities in connectivity. Although many such algorithms exist, detection's No Free Lunch theorem implies that no algorithm can be optimal across all inputs. However, little known practice about how different over or underfit to real networks, reliably assess behavior algorithms. Here, we present broad investigation and underfitting 16 state-of-the-art...

In recent years, critics of online platforms have raised concerns about the ability recommendation algorithms to amplify problematic content, with potentially radicalizing consequences. However, attempts evaluate effect recommenders suffered from a lack appropriate counterfactuals—what user would viewed in absence algorithmic recommendations—and hence cannot disentangle effects algorithm user’s intentions. Here we propose method that call “counterfactual bots” causally estimate role...

We study the fundamental limits on learning latent community structure in dynamic networks. Specifically, we stochastic block models where nodes change their membership over time, but edges are generated independently at each time step. In this setting (which is a special case of several existing models), able to derive detectability threshold exactly, as function rate and strength communities. Below threshold, claim that no algorithm can identify communities better than chance. then give...

Cyberbullying has emerged as an important and growing social problem, wherein people use online networks mobile phones to bully victims with offensive text, images, audio video on a 24/7 basis. This paper studies negative user behavior in the Ask.fm network, popular new site that led many cases of cyberbullying, some leading suicidal behavior.We examine occurrence words Ask.fm's question+answer profiles along network "likes" questions+answers. We also properties users "cutting" this network.

Link prediction algorithms are indispensable tools in many scientific applications by speeding up network data collection and imputing missing connections. However, systems, links change over time it remains unclear how to optimally exploit such temporal information for link predictions networks. Here, we show that topological features, addition having high computational cost, less accurate than sequentially stacked static features. This sequential stacking method uses 41 features avoid...

Cancer incidents are increasingly rising. The quality of care received by cancer patients and the guarantee their satisfaction no longer limited to clinical services. Oncology nurses can offer an important role in enhancing level patient through emotional labor this regard. It aims explore oncology nurses' perceptions regarding concept individual organizational consequences labor. present qualitative research was carried out based on a phenomenological approach. We sampled 18 working at...

Cyberbullying has emerged as an important and growing social problem, wherein people use online networks mobile phones to bully victims with offensive text, images, audio video on a 24/7 basis. This paper studies negative user behavior in the Ask.fm network, popular new site that led many cases of cyberbullying, some leading suicidal behavior. We examine occurrence words Ask.fm's question+answer profiles along network likes questions+answers. also properties users cutting this network.

Real-world network datasets are typically obtained in ways that fail to capture all edges. The patterns of missing data often non-uniform as they reflect biases and other shortcomings different collection methods. Nevertheless, uniform is a common assumption made when no additional information available about the underlying missing-edge pattern, link prediction methods frequently tested against uniformly To investigate impact on accuracy, we employ 9 algorithms from 4 families analyze 20...

Today's densely instrumented world offers tremendous opportunities for continuous acquisition and analysis of multimodal sensor data, providing temporal characterization individual behaviors. Is it possible to efficiently couple such rich data with predictive modeling techniques provide contextual insightful assessments behavior? Such is noisy, incomplete, collected from multiple sources, each a different resolution dimensionality. In addition, longitudinal studies tend examine number...

Negative or antagonistic relationships are common in human social networks, but they less often studied than positive friendly relationships. The existence of a capacity to have and track ties raises the possibility that may serve useful function groups. Here, we analyze empirical data gathered from 24,770 22,513 individuals 176 rural villages Honduras two survey waves 2.5 y apart order evaluate possible relevance for broader network phenomena. We find small-world effect is more significant...

The "friendship paradox" of social networks states that, on average, "your friends have more than you do". Here, we theoretically and empirically explore a related overlooked paradox refer to as the "enmity paradox". We use empirical data from 24,678 people living in 176 villages rural Honduras. show for real negative undirected network (created by symmetrizing antagonistic interactions), exists it does positive world. Specifically, person's enemies enemies, person does. Furthermore, mixed...

Real-world network datasets are typically obtained in ways that fail to capture all links, and there many different non-uniform which real data might be missing. Nevertheless, uniform missing is a common assumption made when no additional information available about the underlying ''missingness function.'' To investigate impact of missingness patterns on link prediction accuracy, we employ 9 algorithms from 4 families analyze 20 functions categorized into 5 groups. By studying 250 real-world...

Human social networks are naturally signed, including both positive and negative ties. However, the instruments for appraising ties less established than those focusing on positive, friendly connections. An improved consensus useable could support comparability, reliability, validity of measures used to map involving This methodological systematic review summarizes classifies "name generators" (NGs) identify ties, applications ascertainment face-to-face relationships in community settings,...

Abstract Link prediction algorithms are indispensable tools in many scientific applications by speeding up network data collection and imputing missing connections. However, systems, links change over time it remains unclear how to optimally exploit such temporal information for link predictions networks. Here, we show that topological features, addition having high computational cost, less accurate than sequentially stacked static features. This sequential stacking method uses 41 features...

In recent years, critics of online platforms have raised concerns about the ability recommendation algorithms to amplify problematic content, with potentially radicalizing consequences. However, attempts evaluate effect recommenders suffered from a lack appropriate counterfactuals -- what user would viewed in absence algorithmic recommendations and hence cannot disentangle effects algorithm user's intentions. Here we propose method that call ``counterfactual bots'' causally estimate role on...

Today's densely instrumented world offers tremendous opportunities for continuous acquisition and analysis of multimodal sensor data providing temporal characterization an individual's behaviors. Is it possible to efficiently couple such rich with predictive modeling techniques provide contextual, insightful assessments individual performance wellbeing? Prediction different aspects human behavior from these noisy, incomplete, heterogeneous bio-behavioral is a challenging problem, beyond...

Cyberbullying has emerged as an important and growing social problem, wherein people use online networks mobile phones to bully victims with offensive text, images, audio video on a 247 basis. This paper studies negative user behavior in the Ask.fm network, popular new site that led many cases of cyberbullying, some leading suicidal behavior.We examine occurrence words Ask.fms question+answer profiles along network likes questions+answers. We also properties users cutting this network.

Although it is under-studied relative to other social media platforms, YouTube arguably the largest and most engaging online consumption platform in world. Recently, YouTube's scale has fueled concerns that users are being radicalized via a combination of biased recommendations ostensibly apolitical anti-woke channels, both which have been claimed direct attention radical political content. Here we test this hypothesis using representative panel more than 300,000 Americans their...