Goals for Winter Vacation

[poorna-kumar]

Minutes of meeting Sat. 17 December:

1. Summary of work done so far.

2. To do:

• Manually inspect articles to check accuracy, quantify error.
• Do lit review.
• Tech: Adjective extraction. Using other cues to determine gender (words like 'her' don't get caught in coref). Think about word2Vec. Databases/API. Subjects vs. sources. How important is a source? Other ways of identifying sources. Annotate important NYT articles and run our aggregate statistics on them.
• Can we use our metrics to come up with a measure of gender diversity? If you control for a story, can you compare how articles are written across organizations? Techmeme.

[viswajithiii]

Updates

• Dec 22: Started annotating NYT page 1 articles. Currently running in parallel. It'll take a few days to run through the entire corpus. There are about ~700 articles marked with page number marked as 1 each month.

  Additional to-dos:

• For NYT articles, we can use Mr. and Mrs. since they use honorifics with all names they mention. Make sure our code can take advantage of this. Maybe even use this as a way to find out how accurate our old methods are.

• Use the names of people mentioned in the NYT corpus as a way to evaluate our methods.