In 2022, researchers in computational social science were eager to employ static word embeddings such as word2vec, although the NLP community had largely moved on from these. While the compelling idea of analogical reasoning in word embeddings had become well known, some of the caveats and potential pitfalls when using these models were not as widely publicized.
To bridge this gap, I presented a lecture on practical considerations for word embeddings at the NLP+CSS 201 lecture series.