Twitter sentiment analysis with a deep neural network: An enhanced approach using user behavioral information

Sentiment analysis on social media such as Twitter has become a very important and challenging task. Due to the characteristics of such data—tweet length, spelling errors, abbreviations, and special characters—the sentiment analysis task in such an environment requires a non-traditional approach. Moreover, social media sentiment analysis is a fundamental problem with many interesting applications. Most current social media sentiment classification methods judge the sentiment polarity primarily according to textual content and neglect other information on these platforms. In this paper, we propose a neural network model that also incorporates user behavioral information within a given document (tweet). The neural network used in this paper is a Convolutional Neural Network (CNN). The system is evaluated on two datasets provided by the SemEval-2016 Workshop. The proposed model outperforms current baseline models (including Naive Bayes and Support Vector Machines), which shows that going beyond the content of a document (tweet) is beneficial in sentiment classification, because it provides the classifier with a deep understanding of the task.     

Spread of Risk Information Through Microblogs: Twitter Users with More Mutual Connections Relay News That is More Dreadful1

In elucidating the spread of risk information through microblogging, it is important to understand the behaviors of numerous average users, in addition to the activities of authorities. We followed the transmission pathways of 10 actual widely spread tweets concerning several risk information topics, including natural disasters, nuclear disasters, and infectious diseases, and we identified the types of risk that affected retweeting by classifying each tweet based on Slovic's risk-perception model. Furthermore, we examined the types of users who did and did not retweet the information. Users with few connections in the form of followers (i.e., people who are following a user) or followees (people a user is following), or with a low ratio of mutual followers within their connections, had a tendency to retweet a large amount of risk information, regardless of the type of risk involved. On the other hand, users with a high ratio of mutual followers exhibited a greater tendency to retweet risk information when it was perceived as dreadful, though they did not retweet risk information much on the whole. These results suggest that there are two mechanisms by which risk information is spread within the Twitter network: information exchange and social sharing of personal reactions.     

Supervisors’ work-related and nonwork information sharing: Integrating research on information sharing,information seeking,and trust using self-disclosure theory

Although significant scholarly attention has been devoted to understanding subordinates’ information seeking from supervisors, researchers have not paid adequate attention to information sharing by supervisors. Moreover, research on supervisors’ information sharing behavior has focused almost exclusively on work-related information sharing, disregarding supervisors’ sharing of information not related to work (e.g., that related to family). Drawing on self-disclosure theory, we argue that supervisors share both work-related and nonwork information with their subordinates and propose that these two forms of information sharing are conceptually distinct. Furthermore, to unravel the role of supervisors’ nonwork information sharing, we develop an interactive model to test how it may be associated with important employee outcomes. We conducted pilot studies using five samples and, through a sixth study, tested the hypothesized model using a four-wave data collection design. This study makes three major contributions to research and theory. First, it integrates information sharing and information seeking literatures. Second, it underlines the importance of supervisors’ nonwork information sharing in organizations by testing its direct and interactive effects. Third, it contributes to theory by presenting trust as a mechanism that links information disclosure in dyads. Results obtained using structural equation modeling generally supported the proposed model.     

差异化变革型领导对知识分享与团队创造力的影响:社会网络机制的解释

差异化变革型领导是团队领导领域的最新议题,现有的研究局限于探讨其社会心理方面的作用机制。本文从社会网络机制出发解释差异化变革型领导对团队知识分享及团队创造力的影响。基于65家企业的225个工作团队样本的结构方程模型分析表明,团队一致性变革型领导正向影响团队交流网络密度从而提高团队知识分享;个体差异性变革型领导正向影响团队成员交流网络密度差异性从而降低团队知识分享;团队知识分享正向影响团队创造力。本文的证据表明,在中国情境下,差异性团队领导行为不利于创造稠密的团队交流网络,以及团队知识分享,从而不利于提升团队创造力水平。     

Cortical dynamics of lateral inhibition: Visual persistence and ISI

Psychophysical studies show that increasing the interstimulus interval (ISI) between two stimuli decreases persistence of the first stimulus. While some researchers account for these results with interactions of transient and sustained inhibition, this paper describes an alternative explanation. In a neural-network model of boundary detection called the boundary contour system, persistence is the result of feedback-generated reverberations. Mechanisms to control these reverberations include lateral inhibition, which computer simulations show allows persistence in the network to qualitatively match the psychophysical data. Additional simulations predict that increasing the duration of the second stimulus should cause persistence to increase with ISI. The model links psychophysical data on visual persistence with computational requirements of spatial vision and properties of cells in visual cortex.     

Computing with connections in visual recognition of origami objects

This paper summarizes our initial foray in tackling Artificial Intelligence problems using a connectionist approach. The particular task chosen was the visual recognition of objects in the Origami world as defined by. The two major questions answered were how to construct a connectionist network to represent and recognize projected line drawings of Origami objects and what advantages such an approach would have. The structure of the resulting connectionist network can be described as a hierarchy of parameter or feature spaces with each node in each of the feature spaces representing a hypothesis about the possible existence of a specific geometric feature of an Origami object. The dynamic behavior of the network is a form of iterative refinement or relaxation whose major characteristic is to prefer more globally interesting interpretations of the input over locally pleasing ones. Examples from the implementation illustrate the system's ability to deal with forms of noise, occlusion and missing information. Other benefits are an inherently parallel approach to vision, limitation of explicit ordering of the search involved in matching model to instance and the elimination of backtracking due to the sharing of partial results as the search progresses. Extensions and problems are also discussed.     

The effect of Twitter exposure on false memory formation

Social media sites such as Facebook and Twitter have increased drastically in popularity. However, information on these sites is not verified and may contain inaccuracies. It is well-established that false information encountered after an event can lead to memory distortion. Therefore, social media may be particularly harmful for autobiographical memory. Here, we tested the effect of Twitter on false memory. We presented participants with a series of images that depicted a story and then presented false information about the images in a scrolling feed that bore either a low or high resemblance to a Twitter feed. Confidence for correct information was similar across the groups, but confidence for suggested information was significantly lower when false information was presented in a Twitter format. We propose that individuals take into account the medium of the message when integrating information into memory.     

Enduring Simples and the Stages They Compose

In this article I introduce a hybrid view of persistence whereby simple objects persist by enduring while composite objects persist by being stage‐related. I first show how, by sharing certain features and not others with the standard (non‐hybrid) views of persistence, this hybrid view navigates two metaphysical problems that have been raised against such standard (non‐hybrid) views. I then consider some implications of the view by addressing a couple of worries that may be raised against it. I conclude that this hybrid view provides an attractive theory, one that deserves further investigation in the debate on persistence.     

Identifying influential segments from word co-occurrence networks using AHP

Identifying important segments in textual data seems to be an important area of research for various applications including topic modelling, trend detection, summarization and event detection. In existing research work, different metrics have been studied to analyse the word co-occurrence network. This research work contributes towards non-semantic and an unsupervised topic identification using the word co-occurrence networks. In this research work, keyphrase have been identified by preserving the lexical sequence using a directed and weighted word co-occurrence network. Further AHP (Analytic Hierarchy Process) model based upon four significant attributes of the word co-occurrence networks have been proposed to rank the keyphrases. Most frequently occurring segment is identified as an influential segment. Experimental results proved high effectiveness of the proposed approach. Results for the First Story Detection, 72 Twitter TDT, synthesized Rio Olympics dataset have been discussed to demonstrate its potential in precisely discovering influential segments.     

Cortical dynamics of visual persistence and temporal integration

In a temporal integration experiment, subjects must integrate two visual stimuli, presented at separate times, to perform an identification task. Many researchers have assumed that the persistence of the leading stimulus determines the ability to integrate the leading and trailing stimuli. However, recent studies of temporal integration have challenged that hypothesis by demonstrating that several theories of persistence are incompatible with data on temporal integration. This paper shows that an account of visual persistence given by a neural network model of preattentive vision, called the boundary contour system, explains data on temporal integration. Computer simulations of the model explain why temporal integration becomes more difficult when the display elements are separated by longer interstimulus intervals or are of longer duration or of higher luminance, or are spatially closer together. The model suggests that different mechanisms underlie the inverse duration effects for leading and for trailing elements. The model further predicts interactions of spatial separation, duration, and luminance of the trailing display.     

Partisan Provocation: The Role of Partisan News Use and Emotional Responses in Political Information Sharing in Social Media

Citizens increasingly rely on social media to consume and disseminate news and information about politics, but the factors that drive political information sharing on these sites are not well understood. This study focused on how online partisan news use influences political information sharing in part because of the distinct negative emotions it arouses in its audience. Using panel survey data collected during the 2012 U.S. presidential election, we found that use of proattitudinal partisan news online is associated with increased anger, but not anxiety, directed at the opposing party's presidential candidate and that anger subsequently facilitated information sharing about the election on social media. The results suggest partisan media may drive online information sharing by generating anger in its audience.     

A Multilevel Model of Team Cultural Diversity and Creativity: The Role of Climate for Inclusion

We developed a multi‐level model to test how team cultural diversity may relate to team‐ and individual‐level creativity, integrating team diversity research and information‐exchange perspective. We proposed that the team climate for inclusion would moderate both the relationship between cultural diversity and team information sharing and between cultural diversity and employee information elaboration. We collected time‐lagged data from 384 members and their leaders within 57 multicultural work teams. The results showed that team cultural diversity was positively related to both team creativity and individual creativity through team information sharing and employee information elaboration, respectively. We also found that the indirect positive relationship with team creativity and individual creativity was stronger for team cultural diversity when climate for inclusion of multicultural work teams was stronger.     

“The Cancer Bond”: Exploring the Formation of Cancer Risk Perception in Families with Lynch Syndrome

This study explores the social context of hereditary cancer risk perception in three families, an African-American family, a Mexican-American family, and a Caucasian family, each with Lynch Syndrome documented by a mismatch repair gene mutation. Communication network assessments measured family communication about cancer experiences and genetic testing information among a total of 26 participants. Participant narratives were evaluated to gain insight into how family cancer experiences and genetic testing information have shaped perceptions of cancer risk. Analysis of communication networks indicated that some families discussed cancer experiences to a greater extent than genetic testing information, and vice-versa. Interviews elucidated that sharing both types of health information led participants to conceptualize linkages among a strong family history of cancer, genetic testing information, and cancer prevention strategies. Understanding how different types of family communication influence the formation of perceived hereditary disease risk may enhance efforts to tailor genetic counseling services for families.     

The effect of frequency of shared features on judgments of semantic similarity

Insight into the structure of conceptual knowledge can be gleaned by examining how statistical regularities in the semantic structure of concepts affect semantic processing. Two similarity judgment experiments revealed that pairs of concepts sharing relatively rare features were judged to be more similar than concepts sharing an equal number of relatively frequent features. Simulations confirmed that these results are consistent with a recurrent connectionist network model of semantic processing in which units corresponding to rare features are activated more quickly and accurately than units corresponding to frequent features. These results support the hypothesis that rare features play a privileged role in the processing of word meanings.     

Hierarchical models for relational event sequences

Interaction within small groups can often be represented as a sequence of events, each event involving a sender and a recipient. Recent methods for modeling network data in continuous time model the rate at which individuals interact conditioned on the previous history of events as well as actor covariates. We present a hierarchical extension for modeling multiple such sequences, facilitating inferences about event-level dynamics and their variation across sequences. The hierarchical approach allows one to share information across sequences in a principled manner—we illustrate the efficacy of such sharing through a set of prediction experiments. After discussing methods for adequacy checking and model selection for this class of models, the method is illustrated with an analysis of high school classroom dynamics from 297 sessions.     

Problems with persistence

A fundamental question in reasoning about change is, what information does a reasoning agent infer about later times from earlier times? I will argue that reasoning about change by an agent is to be modeled in terms of the persistence of the agent's beliefs over time rather than the persistence of truth and that such persistence is explained by pragmatic factors about how agents acquire information from other agents rather than by general principles of persistence about states of the world. AI accounts of persistence have focused on ‘closed world’ examples of change, in which the agent believes that the truth of a proposition is unaltered so long as he or she has no evidence that it has been changed. AI principles of persistence seem plausible in a closed world where one assumes the agent knows everything that is happening. If one drops the assumption of omniscience, however, the analysis of persistence is implausible. To get a good account of persistence and reasoning about change, I argue we should examine ‘open world’ examples of change, in which the agent is ignorant of some of the changes occurring in the world. In open world examples of change, persistence must be formulated, I argue, as a pragmatic principle about the persistence of beliefs. After elaborating my criticisms of current accounts of persistence, I examine how such pragmatic principles fare with the notorious examples of reasoning about action that have collectively characterized the so-called frame problem.     

The Internet is a scary place: How does evidence source and examinee race or ethnicity influence determinations of threat?

Violent rhetoric online is becoming increasingly relevant to the practice of forensic mental health assessment as examinee's virtual lives may transform into real-world acts of violence. With the rise of a diverse subculture of violent online communities, the aim of the present study was to inform how concerns with online sources of collateral data and racial/ethnic biases may influence determinations of violence potential. Using an experimental design, jury-eligible participants (N = 278) and forensic mental health experts (N = 78) were presented with mock Twitter (now referred to as X) posts that varied by data source (i.e., how information was accessed) and the examinee's race/ethnicity. Results showed no differences in participants' ratings of data credibility, how much weight they would place on the posts in a threat assessment, or how likely the examinee was to act violently against his intended target. Implications regarding the interpretation of social media evidence, relevant limitations, and future research are discussed.     

Competence and the maintenance of behavior

The overjustification effect is manifested by decreased interest in an attractive activity following removal of a reward made contingent upon participation in the activity. Evidence is reviewed which suggests that rewards conveying information that the person who is performing the activity is competent cause the behavior to become insulated from the overjustification effect. This maintenance of behavior resulting from competence information is also shown to be a powerful factor in increasing the transfer and posttherapy persistence of behavior in both behavioral and traditional therapies.I wish to thank Pamela Sheridan, Tom Eddy, Larrie Hutton, and Jon Aronoff for their help during the course of my research on this topic. I am also grateful to Sylvia Snavely for sharing her helpful ideas with me.     

Adversarial attacks against profile HMM website fingerprinting detection model

People are accustomed to using an anonymous network to protect their private information. The Profile HMM (Hidden Markov Model) Website Fingerprinting Detection algorithm can detect the website that the data stream accesses by pattern matching the captured data traffic. This makes the anonymous network lose its effect. In order to bypass the detection of this model, we propose a method based on genetic algorithm to generate adversarial samples. By migrating the problem of adversarial samples in deep learning, our approach is used for the broader machine learning detection model to do traffic confusion, and then achieves the purpose of bypassing the Profile HMM model detection. The key challenge is how to construct a suitable fitness function to generate an effective adversarial sample at minimal cost. The experimental results show that the success rate of our traffic confusion method is as high as 97%. At the same time, we only need to add less perturbation traffic than the traditional traffic confusion method.     

Unsupervised discovery of nonlinear structure using contrastive backpropagation

We describe a way of modeling high-dimensional data vectors by using an unsupervised, nonlinear, multilayer neural network in which the activity of each neuron-like unit makes an additive contribution to a global energy score that indicates how surprised the network is by the data vector. The connection weights that determine how the activity of each unit depends on the activities in earlier layers are learned by minimizing the energy assigned to data vectors that are actually observed and maximizing the energy assigned to "confabulations" that are generated by perturbing an observed data vector in a direction that decreases its energy under the current model.