Abstract
Technology’s prevalence in society today relies as much on hype as it does on utility,
with certain innovations gaining and losing traction solely based on the social perspective of that
technology. My theses are related by looking at two different realizations of social hype within
technology: Machine Learning and Makerspaces. Both of these technologies have been put on a
pedestal as unique ways to solve common problems, but my approach to understanding both
differ. The technical thesis involves an in-depth application of Natural Language Processing to
solve the novel problem of detecting and preventing spam, while the STS research focuses on
specifically separating hype from fact by providing an unbiased way to measure the efficacy of
Makerspaces as an educational tool. This report looks at both of these technologies separately,
with the technical thesis applying the technology, and the STS research dissecting the
technology.
The technical thesis focuses on two aspects of detecting and preventing spam using a
novel Machine Learning technique referred to as Natural Language Processing. Initially, work
was done on a chat bot placed in the undermarket web to communicate with e-commerce
miscreants and extract useful intel. For this project, Natural Language Processing was used to
improve the machine’s understanding of the conversation with the miscreant, thus enabling it
with richer information. Utilizing a new NLP framework, we were able to substantially improve
the understanding of textual information and flow of conversation for the chat bot. Additionally,
in a separate project, NLP is being used to build a developer’s reputation and predict whether
that user is commiting malicious code to GitHub. We are currently extracting textual information
from commited code and vectorizing to produce more features for the prediction.
The STS thesis investigates the efficiency of Makerspaces as a learning tool, attempting
to separate what is fact and what is hype. As many schools begin to adopt Makerspaces, the need
to define the metrics by which to evaluate these collaborative spaces increases. In this research, I
discuss the existing methods of evaluation from both an educational and social lens.
Furthermore, I produce metrics that additionally account for the resources used, the opportunity
cost of those resources, and the specific benefits reaped. Through case studies of existing
Makerspaces, it was found that successful makerspaces were coupled with an existing
curriculum and had strong support from patrons. The research recommends an amended
evaluation metric for Makerspaces, as well as key factors that contribute to the efficiency of
these spaces. The research is preliminary yet substantive and further research into Makerspaces
could be grounded in what is found here.
