[joeroot]

I like the idea, however I'd disagree with the heuristic used. DS9 was really the first Star Trek series to introduce story arcs, and as such, skipping the wrong episodes leads to a confusing experience. I'd argue that in order to avoid missing out on important chunks of the arc, hand curation is really the only viable option (maybe text-clustering of plot narratives might work?).

Personally, as a huge DS9 fan, I think that you should watch every episode. Each episode adds colour and depth to the series' characters, and in my opinion makes it the most rewarding Star Trek series.

[danielweber]

I tend to be of the "watch it all" camp, since even mediocre episodes make the excellent episodes much more rewarding. And you don't know which is which going in.

That said, "Profit And Lace" has no redeeming qualities whatsoever.

[delecti]

As bad as it is (and it really was retched), even "Profit and Lace" plays into the larger storyline of Ferengi women gaining civil rights.

[tessa_t]

I tried to keep story arcs in mind and just skip the "filler" episodes. Also skipped the lame story arcs, like those "the circle" people in S1/S2

[joeroot]

I see, apologies! I thought you had pulled in all episodes with a score > 7.

I wonder how easily arcs can be identified. I'll try running the transcripts through a topic model this evening.

[jkaunisv1]

You say that so casually. As someone with no experience in NLP, do you just have topic model algo's or a toolkit lying around? Going on your website this isn't your field of expertise, have you worked on this stuff in the past or just a hobby?

[joeroot]

Enough to get by! I worked on this throughout university, and now at my startup. I was slightly blasé however! The corpus is tiny (173 episodes: http://www.chakoteya.net/ds9/episodes.htm), so a topic model is unlikely to yield anything valuable. There are probably around 10-15 arcs, and simple clustering could be better -- but this is purely hypothetical. In this case, it's simply curiosity.

If you're interested in tools, Mallet (http://mallet.cs.umass.edu/) is a fairly good place to start, and the original LDA paper by Blei, Ng & Jordan (http://machinelearning.wustl.edu/mlpapers/paper_files/BleiNJ...) is a great academic starting point.

[jkaunisv1]

Cool, thanks!

[tessa_t]

Most of the arc episodes are labeled as part of an arc, like "When it rains... (5)"