Date of Award
Master of Science
The advent of the Internet has yielded exciting new opportunities for the collection of large amounts of structured and unstructured social scientific data. This thesis describes two such methods for harvesting data from websites and web services: web-scraping and connecting to an application programming interface (API). I describe the development and implementation of tools for each of these methods. In my review of the two related, yet distinct data collection methods, I provide concrete examples of each. To illustrate the first method, ‘scraping’ data from publicly available data repositories (specifically the Google Books Ngram Corpus), I developed a tool and made it available to the public on a web site. The Google Books Ngram Corpus contains groups of words used in millions of books that were digitized and catalogued. The corpus has been made available for public use, but in current form, accessing the data is tedious, time consuming and error prone. For the second method, utilizing an API from a web service (specifically the Twitter Streaming API), I used a code library and the R programming language to develop a program that connects to the Twitter API to collect public posts known as tweets. I review prior studies that have used these data, after which, I report results from a case study involving references to countries. The relative prestige of nations are compared based on the frequency of mentions in English literature and mentions in tweets.
Willers, Joel, "Methods for extracting data from the Internet" (2017). Graduate Theses and Dissertations. 15643.