Issue 1588 Improve Made The List insights. Adds list word frequency counts #1823

Closed
wants to merge 1 commit into
from

Conversation

Projects
None yet
3 participants
Contributor

gbrady92 commented Feb 23, 2014

Hey Gina,
I've completed the improvements on the "Made the List" insight. I've added the functionality to count the frequency of the words in the list titles. I have then added the top 4 words to the insight for the user to read. Is there any changes you would like me to make or are you happy with these improvements ?

Thanks,
Gareth

Owner

anildash commented May 29, 2014

Gareth, I'll defer to Gina on reviewing the actual code, but I wanted to offer some feedback on this PR overall.

  • First, the overall concept here is great! Any time we can be extracting data and summarizing it in efficient ways for the user, that's right on target.
  • For this particular insight, this may not be a frequent use case for most ThinkUp users. People tend to be added to a handful of lists in any given time period, even if they have enormous follower counts.
  • There do tend to be clusters or clumping of descriptors when people are added to lists. (Most of the lists I'm on describe me as "tech", for instance.) That means counts as a concept may be useful, but perhaps mostly in the context of all lists a person is on, rather than just recently-added ones.
  • Let's get back to first principles: How do we want the user to feel when they see this? Our phrasing for the insight overall is to ask the user if they think those list names are a good descriptor. So the big leap here is that we recognized that people use lists almost as "tags" to describe people, and that the accuracy (or flattery) of those tags might be something a user would respond to in a really emotional or evocative way. If the most common tag/list name for me in a given week is "tech" (which it will be for me 90% of the time), I'm not sure I have a different emotion between seeing it was used 3 times and 4 times.

The capability to summarize top words, along with counts, is really powerful. But perhaps we should think a bit about the best places to use that power. Might we take this same code and focus it instead on something like which words a person used in their posts, or which words were most common in posts that inspired the user to reply?

Contributor

gbrady92 commented May 30, 2014

Hey Anil,
Thanks for the feedback on the insight's improvements, this is actually the first bit of code I worked on a few months back so I was expecting there to be some room for improvement.

I will look into how to increase the scope of the insight, and try out different ways to display the information within the insight.

I like the idea of using counts in different situations like you suggested and will look into creating some insights to take advantage of this. At the moment I am working on another insight (distance shared) but once that is done I'll turn my attention onto this one.

Thanks again for the feedback.

Contributor

cdmoyer commented May 30, 2014

A thought I had was that for this insight, it could pull in and look at all your lists when it triggers. So:

@cdmoyer was just adde to a new list: "geeks" bringing him up to 18 lists. 50% of @cdmoyer's list are now about "geekery" or "tech".

On May 30, 2014, at 11:55 AM, gbrady92 notifications@github.com wrote:

Hey Anil,
Thanks for the feedback on the insight's improvements, this is actually the first bit of code I worked on a few months back so I was expecting there to be some room for improvement.

I will look into how to increase the scope of the insight, and try out different ways to display the information within the insight.

I like the idea of using counts in different situations like you suggested and will look into creating some insights to take advantage of this. At the moment I am working on another insight (distance shared) but once that is done I'll turn my attention onto this one.

Thanks again for the feedback.


Reply to this email directly or view it on GitHub.

Contributor

cdmoyer commented Jul 11, 2014

I'm going to close the pull request for housekeeping. Feel free to open a new one if you circle back around and make the discussed chanes.

cdmoyer closed this Jul 11, 2014

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment