FreeTime mode on the Amazon Kindle will provide you with reading statistics. You can find the raw data as an SQLite file under system > freetime > freetime.db.

Ok, here I'm getting a bit more worried. It's not that I don't think this is helpful. But I do think it's skipping some possible better, more human solutions.

One concern: the premise here is that comprehension struggles are mostly questions requiring answers rather than discursive situations requiring more interaction. A second related concern: is the ultimate goal of "learning" to get the answer or to acquire facility with that discursive process? (Answer: the latter.)

I think simple social annotation, perhaps backed by some AI, could go a long way here. Allow students to ask questions, answer each others questions, and surface those questions and answers in a useful way to teachers...

S3Object is a pointer to the data object.

Provides an InputStream to read the data.

p. 6

Retrieval methods designed for small databases decline rapidly in effectiveness as collections grow...

This is an interesting point that is missed in the Distant reading controversies: its all very well to say that you prefer close reading, but close reading doesn't scale--or rather the methodologies used to decide what to close read were developed when big data didn't exist. How to you combine that when you can read everything. I.e. You close read Dickins because he's what survived the 19th C as being worth reading. But now, if we could recover everything from the 19th C how do you justify methodologically not looking more widely?