I focus on connections between members (who's talking to who, who's talking to the most people, etc)
I also track who is helping others, in which channels, and on what topics.
Lastly, I track welcoming messages (Hi, Hello, Welcome, etc) and thankful messages (Thanks, Thank you, thx, etc) to see the overall mood of the community.
Love that you track thanks for sentiment! Are there other key phrases you like to track for negative sentiment and, if so, how do you balance that with the positive (if you can't speak to this I totally understand)?
In my experience it's too hard to try and gauge positive/negative sentiment in comments. Even the best AI I've see struggles with being accurate enough to be useful. So I track these two specific kinds of messages, which reliably have some specific keywords in them.
But I do use keywords to tag conversations by topic, like a specific piece of software or initiative I'm running. Here's just some of what I have right now:
FWIW, I created a Tag to try and track negative sentiment using keywords "sucks, horrible, terrible, awful, very bad, regret".
Overall, it didn't match many comments. Some were false-positives (somebody calling their own example "terrible", for instance), but it did identify a few legitimately negative comments.
Discussion (10)
I focus on connections between members (who's talking to who, who's talking to the most people, etc)
I also track who is helping others, in which channels, and on what topics.
Lastly, I track welcoming messages (Hi, Hello, Welcome, etc) and thankful messages (Thanks, Thank you, thx, etc) to see the overall mood of the community.
Love that you track thanks for sentiment! Are there other key phrases you like to track for negative sentiment and, if so, how do you balance that with the positive (if you can't speak to this I totally understand)?
In my experience it's too hard to try and gauge positive/negative sentiment in comments. Even the best AI I've see struggles with being accurate enough to be useful. So I track these two specific kinds of messages, which reliably have some specific keywords in them.
But I do use keywords to tag conversations by topic, like a specific piece of software or initiative I'm running. Here's just some of what I have right now:
FWIW, I created a Tag to try and track negative sentiment using keywords "sucks, horrible, terrible, awful, very bad, regret".
Overall, it didn't match many comments. Some were false-positives (somebody calling their own example "terrible", for instance), but it did identify a few legitimately negative comments.
Do you have a protocol for action if you see something moving in the wrong direction?
I do! I wrote them up here: docs.savannahhq.com/insights/conne...
I'm currently in the middle of implementing two different changes in two different parts of my community based on how my connection graph looks.
How did you make this connections graph? It's awesome!
They're provided automatically by savannahhq.com
I believe you already have an account on there
I am so excited about those connections graphs! Thanks for sharing :D
We're really focused on overall growth and weekly active users