“The Enemy Among Us”

Recently, numerous approaches have emerged in the social sciences to exploit the opportunities made possible by the vast amounts of data generated by online social networks (OSNs). Having access to information about users on such a scale opens up a range of possibilities—from predicting individuals’ demographics and health status to their beliefs and political opinions—all without the limitations associated with often slow and expensive paper-based polls. A question that remains to be satisfactorily addressed, however, is how demography is represented in OSN content—that is, what are the relevant aspects that constitute detectable large-scale patterns in language? Here, we study language use in the United States using a corpus of text compiled from over half a billion geotagged messages from the online microblogging platform Twitter. Our intention is to reveal the most important spatial patterns in language use in an unsupervised manner and relate them to demographics. Our approach is based on Latent Semantic Analysis augmented with the Robust Principal Component Analysis methodology, which permits identification of the data’s main sources of variation with an automatic filtering of noise and outliers without influencing results by a priori assumptions. We find spatially correlated patterns that can be interpreted based on the words associated with them. The main language features can be related to slang use, urbanization, travel, religion and ethnicity, the patterns of which are shown to correlate plausibly with traditional census data. Apart from the standard measure of linear correlation, some relations seem to be better explained by Boolean implications, suggesting a threshold-like behaviour where demographic variables influence the users’ word use. Our findings validate the concept of demography being represented in OSN language use and show that the traits observed are inherently present in the word frequencies without any previous assumptions about the dataset. They therefore could form the basis of further research focusing on the evaluation of demographic data estimation from other big data sources, or on the dynamical processes that result in the patterns identified here.

This paper describes the design and evaluation of NFriendConnector, a prototype application that allows for better integration between online and offline social networks. Online social networks are currently used to maintain and strengthen existing real-life social connections, rather than establishing ties that exist only online. However, users incur significant time and search related costs in replicating a naturally occurring social interaction using a social networking site (SNS). Therefore, there exists a gap between initiating social contact in real-life versus initiating social contact via an online social network. Using the design science paradigm, our research addresses this gap by introducing NFriendConnector. This application allows users to map their offline interactions, as and when they take place, onto their SNS presence, therefore making it possible to complement offline social interactions with SNS profile information. The prototype is implemented using Near Field Communication (NFC)-enabled mobile phones and Facebook. We evaluate the prototype in an experimental setting using expectation confirmation theory (ECT) as the theoretical framework. Findings show that NFriendConnector was able to satisfy users, therefore indicating a successful design exercise. We discuss the implications of this research in the context of current developments in online social networking.

In content based image retrieval (CBIR) process, every image has been represented in a compact set of local visual features i.e. color, texture, and/or shape of images. This set of local visual features is known as feature vector. In the CBIR process, feature vectors of images have been used to represent or to identify similar images in adequate way. As a result, feature vector construction has always been considered as an important issue since it must reflect proper image semantics using minimal amount of data. The proposed CBIR scheme is based on the combination of color and texture features. In this work initially, we have converted the given RGB image into HSV color image. Subsequently, we have considered H (hue), S (saturation), and V (intensity) components for extraction of visual image features. The texture features have been extracted from the V component of the image using 2D dual-tree complex wavelet transform (2D DT-CWT) where it analyzes the textural patterns in six different directions i.e. ±15\(^{\circ }\), ±45\(^{\circ }\), and ±75\(^{\circ }\). At the same time, we have computed the probability histograms of H and S components of the image respectively and subsequently those are divided into non-uniform bins based on cumulative probability for extraction of color based features. So, in this work both the color and texture features have been extracted simultaneously. Finally, the obtained features have been concatenated to attain the final feature vector and same is considered in image retrieval process. We have tested the novelty and performance of the proposed work in two Corel, two objects, and, a texture image datasets. The experimental results reveal the acceptable retrieval performances for different types of datasets.

Charting the use of inclusive language in medical school diversity mission statements: A national study.

The study explores the practice, meaning and significance of invective and insults in political discourse among Ghanaians. It examines the use of such language forms against people of higher social status in online discourse in Ghana, a country where it is traditionally unacceptable to openly criticize the elderly and authority figures, let alone insult them. Data was sourced from GhanaWeb [www.ghanaweb.com], the most patronized online news website in Ghana that ensures anonymous citizen participation in issues of national interest. The data involved invective and insult-filled online reader comments on news reports about the presidential candidates and the electoral commissioner involved in the 2016 general election in Ghana. Metapragmatic comments (reader reactions) which portray Ghanaian sociocultural perspectives on the use of abusive language against these individuals were also gathered. These two sets of comments were used to build the Corpus of GhanaWeb Comments on Ghana’s 2016 Election (CGCGE16). In addition to CGCGE16, the data for the study included comments from social media platforms, native speakers’ intuitions and the views of Ghanaian anthropological linguists and cultural specialists obtained from interviews, and evidence from scholarly works. The qualitative meaning analysis method is used along with the quantitative discourse analysis method. The study adopts an ethnopragmatic approach, combining corpus-based searches with semantic explications and cultural scripts, following the principles of the Natural Semantic Metalanguage (NSM) approach (Wierzbicka, 1992, 1996; Goddard & Wierzbicka, 2014). The findings indicate that most of the comments in CGCGE16 are personal attacks that violate the Ghanaian communicative norms towards the elderly and people in authority and, furthermore, do not conform to the expected displays of verbal respect. Instead of expressing their views on on-going socio-political issues, the commenters often draw on past socio-political occurrences, their public and sociocultural knowledge to engage in negative emotional characterizations and evaluations of the presidential candidates and the electoral commissioner, thereby, showing disapproval of them. By means of name-calling expressions, derogatory adjectives, and derogatory interjections, these commenters attack the intellectual abilities, moral behaviour, social competence, and physical characteristics of their targets. This negative communicative behaviour is attributed to the fact that on GhanaWeb, like on many online communities, interactants are not obliged to uphold communicative norms. The anonymity of GhanaWeb shields commenters from the repercussions such speech behaviours would otherwise necessitate. It is also attributed to the right to freedom of speech, the repeal of the criminal libel law, and the fact that traditionally, there are some specific settings that allow the use of insults against authority. The study demonstrates that although the practice of using insults against political leaders in online commentary appears to be “normal” in the online context, it is a transgression when one considers the Ghanaian sociocultural norms of communication. Consistent with this, although the interaction on GhanaWeb is asynchronous, some commenters play the role of regulatory figures and traditional gatekeepers of appropriate communicative behaviour by posting disapproving metapragmatic comments. The study identifies many ways in which interactions on GhanaWeb are culturally shaped. These include first, commenters do not only rely on English slang terms and swearwords, they also employ some local language insulting expressions (e.g. kwasea ‘fool’, yaka gbɔmɔ ‘worthless person’) due to the ‘weight’ such expressions carry for Ghanaians. Secondly, the metapragmatic comments do not employ certain key terms such as “offensiveness” that can be found in the literature on online studies related to the speakers of Inner Circle Englishes. Instead, they highlight the clear position of the Ghanaian culture on respect for elders and people in authority. These metapragmatic comments also stress the fact that the exchange of invective and insults in politics can result in violence which may have adverse effects on the development of the nation. Overall, the study is inclined towards the view that instead of generalizing the results of studies on abusive language use in online interactions, there is a need to draw out what distinguishes Ghanaian online interactants from, let’s say, Western countries, Asian countries, African countries, or from interactants elsewhere. The key claim is that understanding the sociocultural backdrop for the speech practices of online interactants is crucial to understanding their language use, as interactants draw on existing communicative repertoires which influence their language choices. Also, the study draws attention to news websites in Ghana as a platform where the use of unmitigated invective and insults against people of higher social status is increasingly becoming the norm, despite the fact it violates Ghanaian communicative values. Thus, this study serves as a springboard for future research into abusive language use in online interactions among Ghanaians.

In the attribute inference problem, we aim to infer users' private attributes (e.g., locations, sexual orientation, and interests) using their public data in online social networks. State-of-the-art methods leverage a user's both public friends and public behaviors (e.g., page likes on Facebook, apps that the user reviewed on Google Play) to infer the user's private attributes. However, these methods suffer from two key limitations: 1) suppose we aim to infer a certain attribute for a target user using a training dataset, they only leverage the labeled users who have the attribute, while ignoring the label information of users who do not have the attribute; 2) they are inefficient because they infer attributes for target users one by one. As a result, they have limited accuracies and applicability in real-world social networks. In this work, we propose AttriInfer, a new method to infer user attributes in online social networks. AttriInfer can leverage both friends and behaviors, as well as the label information of training users who have an attribute and who do not have the attribute. Specifically, we model a social network as a pairwise Markov Random Field (pMRF). Given a training dataset, which consists of some users who have a certain attribute and some users who do not have a certain attribute, we compute the posterior probability that a target user has the attribute and use the posterior probability to infer attributes. In the basic version of AttriInfer, we use Loopy Belief Propagation (LBP) to compute the posterior probability. However, LBP is not scalable to very large-scale real-world social networks and not guaranteed to converge. Therefore, we further optimize LBP to be scalable and guaranteed to converge. We evaluated our method and compare it with state-of-the-art methods using a real-world Google+ dataset with 5.7M users. Our results demonstrate that our method substantially outperforms state-of-the-art methods in terms of both accuracy and efficiency.

Previous investigators have demonstrated that gender differences exist in the use of sexual language in various interpersonal contexts. However, sexual orientation has not been addressed in any study regarding sexual language usage. This investigation examined the use of sexual language for female and male in different interpersonal contexts based upon sexual orientation. Respondents were 440 university students: 202 heterosexual female, 30 lesbian, 173 heterosexual male, and 35 gay male. Findings indicate that differences in sexual language usage in interpersonal contexts is significant for gender as well as for sexual orientation. Conjecture is made for the implications this may have on intimate relationships.

Trust relationship prediction among people provides valuable supports for decision making, information dissemination, and product promotion in online social networks. Network embedding has achieved promising performance for link prediction by learning node representations that encode intrinsic network structures. However, most of the existing network embedding solutions cannot effectively capture the properties of a trust network that has directed edges and nodes with in/out links. Furthermore, there usually exist rich user attributes in trust networks, such as ratings, reviews, and the rated/reviewed items, which may exert significant impacts on the formation of trust relationships. It is still lacking a network embedding-based method that can adequately integrate these properties for trust prediction. In this work, we develop an AtNE-Trust model to address these issues. We firstly capture user embedding from both the trust network structures and user attributes. Then we design a deep multi-view representation learning module to further mine and fuse the obtained user embedding. Finally, a trust evaluation module is developed to predict the trust relationships between users. Representation learning and trust evaluation are optimized together to capture high-quality user embedding and make accurate predictions simultaneously. A set of experiments against the real-world datasets demonstrates the effectiveness of the proposed approach.

In recent years, due to the booming development of online social networks, fake news for various commercial and political purposes has been appearing in large numbers and widespread in the online world. With deceptive words, online social network users can get infected by these online fake news easily, which has brought about tremendous effects on the offline society already. An important goal in improving the trustworthiness of information in online social networks is to identify the fake news timely. This paper aims at investigating the principles, methodologies and algorithms for detecting fake news articles, creators and subjects from online social networks and evaluating the corresponding performance. This paper addresses the challenges introduced by the unknown characteristics of fake news and diverse connections among news articles, creators and subjects. This paper introduces a novel gated graph neural network, namely FAKEDETECTOR. Based on a set of explicit and latent features extracted from the textual information, FAKEDETECTOR builds a deep diffusive network model to learn the representations of news articles, creators and subjects simultaneously. Extensive experiments have been done on a real-world fake news dataset to compare FAKEDETECTOR with several state-of-the-art models, and the experimental results are provided in the full-version of this paper at [13].

The Command and Control (C&C) channel of modern botnets is migrating from traditional centralized solutions (such as the ones based on Internet Relay Chat and Hyper Text Transfer Protocol), towards new decentralized approaches. As an example, in order to conceal their traffic and avoid blacklisting mechanisms, recent C&C channels use peer-to-peer networks or abuse popular Online Social Networks (OSNs). A key reason for this paradigm shift is that current detection systems become quite effective in detecting centralized C&C. In this paper we propose ELISA (Elusive Social Army), a botnet that conceals C&C information using OSNs accounts of unaware users. In particular, ELISA exploits in a opportunistic way the messages that users exchange through the OSN. Furthermore, we provide our prototype implementation of ELISA. We show that several popular social networks can be maliciously exploited to run this type of botnet, and we discuss why current traffic analysis systems cannot detect ELISA. Finally, we run a thorough set of experiments that confirm the feasibility of our proposal. We have no evidence of any real-world botnets that use our technique to create C&C channels. However, we believe that finding out in advance potential new types of botnets will help to prevent possible future malevolent applications.

Modeling the spread of influence in online social networks is important for predicting the influence of individualsand better understanding many scenarios in social networks, such as the influence maximization problem. The previous work on modeling the spread of influence makes the assumption that the statuses of nodes in a network are independent of each other,which is apparently not correct for social networks. The goal of this work is to derive an accurate mathematical model to characterize the spread of influence for the independent cascade diffusion process in online social networks. Specifically, we apply the susceptible-infected-recovered epidemic model from epidemiology to characterize the independent cascade diffusion processand derive a general mathematical framework. To approximate the complex spatial dependence among nodes in a network, we propose a Markov model to predict the spread of influence. Through the extensive simulation study over several generated topologies and a real coauthorship network,we show that our designed Markov model has much better performance than the existing independent model in predicting the influence of individuals in online social networks.

Graduate student cognition during information retrieval using the World Wide Web: a pilot study

Science epistemology, or beliefs about what it means to do science and how science knowledge is generated, is an integral part of authentic science inquiry. Although the development of a sophisticated science epistemology is critical for attaining science literacy, epistemology remains an elusive construct to precisely and quantitatively evaluate. Previous work has suggested that analysis of student practices in science inquiry, such as their use of language, may be reflective of their underlying epistemologies. Here we describe the usage of a learning analytics tool, TAALES, and keyness analysis to analyze the concluding statements made by students at the end of a computer-based authentic science inquiry experience. Preliminary results indicate that linguistic analysis reveals differences in domain-general lexical sophistication and in domain-specific verb usage that are consistent with the expertise level of the participant. For example, experts tend to use more hedging language such as "may" and "support" during conclusions whereas novices use stronger language such as "cause." Using these differences, a simple, rule-based prediction algorithm with LOOCV achieved prediction accuracies of greater than 80%. These data underscore the potential for the use of learning analytics in simulated authentic inquiry to provide a novel and valuable method of assessing inquiry practices and related epistemologies.

SentiHealth-Cancer: A sentiment analysis tool to help detecting mood of patients in online social networks

The continued growth of Internet traffic associated with the popularity of online social networks (OSNs) raises the possibility that knowledge of the underlying social graph connecting users may be exploited to enhance content distribution and delivery. In our previous work, we found that partitioning OSN users and their data based on their social graph, and using the partitioning as basis for content placement decisions across cache servers yield significant improvements over randomly-partitioned implementations. In this paper we go beyond content placement and use the same partitioning scheme to push updates across distributed caches to maintain consistency in OSN state. To avoid making singular assumptions on the degree to which user interaction patterns are influenced either by social considerations or content popularity, we parameterized our experiments to determine the impact of various interaction mixes on different cache designs. We found that in the new hybrid scheme that combines socially-aware cooperative content placement, on one hand, with social graph-driven proactive content delivery on the other, significant performance benefit continues to be driven by the former rather than the latter. Consequently, the hybrid scheme is likewise sensitive to the granularity of graph partitioning, which in turn dictates cache sizes and the number of cache servers.