(The following may be later integrated into my completed project.)

To complete the NULab Certificate, I have chosen to utilize the power of word embedding models for a project on the Foreign Relations of the United States series. The project initially was based around a basic gender analysis on two regions over the period of the early Cold War. Operating under the assumption that the corpus was a complete representation of all foreign relations documents concerning the area and timeframe, I began to investigate the corpus through basic term searches. Directing my search toward strongly gendered vectors, either male or female, two very different results were produced from each corpus. In the European corpus, words associated with the terms “woman+girl” resembled roles in relation to men, such as “daughter” and “wife”, but also terms of agriculture such as “mule” and “lamb”. From the corpus regarding Latin America in the same period with the same search terms, associated words included “jail”, “hanged”, “traitor”, “killed”, and more. These two queries included the same parameters (Threads: 3; Vectors: 100; Window: 10; Iterations: 10; Negative Sample: 5).