Social media.

One of the de ning characteristics of the modern Internet is its massive connectedness,
with information and human connection simply a few clicks away. Social
media and online retailers have revolutionized how we communicate and purchase
goods or services. User generated content on the web, through social media, plays
a large role in modern society; Twitter has been in the forefront of political discourse,
with politicians choosing it as their platform for disseminating information,
while websites like Amazon and Yelp allow users to share their opinions on products
via online reviews. The information available through these platforms can provide
insight into a host of relevant topics through the process of machine learning. Speci -
cally, this process involves text mining for sentiment analysis, which is an application
domain of machine learning involving the extraction of emotion from text.
Unfortunately, there are still those with malicious intent and with the changes
to how we communicate and conduct business, comes changes to their malicious practices.
Social bots and fake reviews plague the web, providing incorrect information
and swaying the opinion of unaware readers. The detection of these false users or
posts from reading the text is di cult, if not impossible, for humans. Fortunately, text mining provides us with methods for the detection of harmful user generated
content.
This dissertation expands the current research in sentiment analysis, fake online
review detection and election prediction. We examine cross-domain sentiment
analysis using tweets and reviews. Novel techniques combining ensemble and feature
selection methods are proposed for the domain of online spam review detection. We
investigate the ability for the Twitter platform to predict the United States 2016 presidential
election. In addition, we determine how social bots in
uence this prediction.

This thesis project seeks to answer the question of how visual rhetoric put forward
in social media content by pro-life and pro-choice organizations may persuade their
audiences’ perspective on abortion. Using Sonja Foss’s guidelines for analysis of visual
rhetoric, I analyze 24 selected examples of Facebook content posted by two pro-life
organizations (Human Coalition and Feminists for Life) and two pro-choice organizations
(Planned Parenthood Action and NARAL Pro-Choice America) in 2017.
My analysis found that the visual rhetoric posted by both organizations on social
media can and does function as a form of visual metonymy. Because of this, these visual
strategies can stand in for more complex arguments in dramatic ways.

Exclusive breastfeeding for the first six months of life has been shown to decrease morbidity and mortality of women and infants. Organizations such as the United Nations Children’s Fund (UNICEF, 2018), American Academy of Pediatrics (AAP, 2012), and the World Health Organization (WHO, 2017a) have universally endorsed exclusive breastfeeding for the first six months of life, and then continuation of breastfeeding for a minimum of one to two years, with only supplementation of other liquid or solid food sources. Breastfeeding rates in the United States have not met the minimum goals set forth by Healthy People 2020 (n.d.). Although 81% of U.S. mothers initiated breastfeeding after the birth of their infant, only 22% of mothers were found to be exclusively breastfeeding at six months postpartum (Centers for Disease Control and Prevention [CDC], 2016a).
This prospective, longitudinal, structural equation modeling study examined millennial-aged, exclusively breastfeeding women within one month postpartum who were followers of at least one of 17 social media breastfeeding support groups. Relationships of the conceptual constructs within Pender’s (1996) revised health promotion model (RHPM); House’s (1981) dimensions of social support; and the added constructs of breastfeeding knowledge, breastfeeding confidence, and breastfeeding attitude were analyzed in an effort to better understand the variables that lead to sustained exclusive breastfeeding to six months.
Data supported the use of the integrated model for breastfeeding women. The normed referenced chi-square (2) of 1.9 (CFI =.94, IFI =.94, NFI =.89, RMSEA =.06, CFI [PCFI] >.5) indicated a good model fit. Additionally, there were statistically significant gains in the confidence, knowledge, and attitude scores from pretest to follow-up at six months. Exclusive breastfeeding to six months was reported to be three times (66%) higher than the U.S. national average (22%) (CDC, 2016a). Future use of the integrated model has great potential to impact public health by the exploration of variables that promote exclusive breastfeeding to six months.

This study used a training package to teach social media safety skills, using
Facebook, to adolescents and young adults diagnosed with autism spectrum disorders in a
small group setting. Participants were taught to decline, block and report when they
received a lure from someone that they do not know. A multiple baseline design across
lures demonstrated the effects of the intervention on participant performance. Results
confirmed an increase in social media safety skills performed by all participants.
Participants were able to maintain this skill set once the training package was removed.
Spontaneous generalization was demonstrated by all participants for some lures.
Generalization of social media safety skills was demonstrated across participants in a
setting where they did not receive instruction. Limitations and implications for future
research are discussed.

How we share our good news with people can have a significant effect on our
lives. Sharing good news on social media sites involves a process called capitalization.
Capitalization has been shown to increase well-being when others provide appropriate
responses in face-to-face interactions. To see if this effect on well-being extends to our
online presence, this study utilized the social media site Facebook to observe if
capitalization predicted well-being and relationship satisfaction. This study used data
collected from 137 participants recruited from an undergraduate participant pool and
from Amazon Mechanical Turk. Consistent with hypotheses, participants who reported
receiving active and constructive responses after sharing a positive event on Facebook
also reported greater personal well-being and relationship satisfaction. Although future
experimental research is needed to establish causality, the current results suggest that the ways in which friends respond to social media posts are associated with personal and
relationship well-being.

Social media posts are used to examine what people experience in their everyday
lives. A new method is developed for assessing the situational characteristics of social
media posts based on the words used in these posts. To accomplish this, machine learning
models are built that accurately approximate the judgments of human raters. This new
method of situational assessment is applied on two of the most popular social media sites:
Twitter and Facebook. Millions of Tweets and Facebook statuses are analyzed. Temporal
patterns of situational experiences are found. Geographic and gender differences in
experience are examined. Relationships between personality and situation experience
were also assessed. Implications of these finding and future applications of this new
method of situational assessment are discussed.

Sentiment analysis of tweets is an application of mining Twitter, and is growing
in popularity as a means of determining public opinion. Machine learning algorithms
are used to perform sentiment analysis; however, data quality issues such as high dimensionality, class imbalance or noise may negatively impact classifier performance.
Machine learning techniques exist for targeting these problems, but have not been
applied to this domain, or have not been studied in detail. In this thesis we discuss
research that has been conducted on tweet sentiment classification, its accompanying
data concerns, and methods of addressing these concerns. We test the impact
of feature selection, data sampling and ensemble techniques in an effort to improve
classifier performance. We also evaluate the combination of feature selection and
ensemble techniques and examine the effects of high dimensionality when combining
multiple types of features. Additionally, we provide strategies and insights for
potential avenues of future work.