Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Change headers (of multiple responses) to questions #45

Closed
berkingurcan opened this issue May 25, 2024 · 13 comments · Fixed by #56
Closed

Change headers (of multiple responses) to questions #45

berkingurcan opened this issue May 25, 2024 · 13 comments · Fixed by #56
Assignees
Labels
approved-for-product enhancement New feature or request PR 56 Survey Bot A bot to summarize and categorize surveys tested!

Comments

@berkingurcan
Copy link
Collaborator

@es92 I think the bug is this right? It needs to be questions instead of RESPONSE 1,2,3,4,5 right?
Screenshot 2024-05-25 at 17 30 38

If not, could you explain the bug?

@berkingurcan
Copy link
Collaborator Author

berkingurcan commented May 27, 2024

@es92
Now I have implemented this in this branch:

  • We can collect multi questions like in Survey Bot
    by entering lines for each question:
Screenshot 2024-05-27 at 15 10 30
  • Then we can answer like this
Screenshot 2024-05-27 at 14 39 28 Screenshot 2024-05-27 at 14 39 46

Note: Can not be added text fields, so I have added Survey description as placeholder. Not looks good but functionally OK.

@es92
Copy link
Collaborator

es92 commented May 27, 2024

Yeah I think its unclear what's going on right on the bot -- we could do this -- but to explain more what is happening:

There are really two survey modes for the bot. Mode 1 is to take all of the text the user submitted, and to make that one datapoint that gets sorted into one topic and one subtopic.

Mode 2 is to allow for multiple pieces of text to be submitted, that each get categorized into its own topic and subtopic.

I was thinking "mode 1" would be for things like sentiment -- eg "what do you think about this", where each submission ends up in its own place in the results. And "mode 2" would be for things like brainstorming or feedback -- eg "what changes would you make to this bot", where each submission is separate and ends up in its own categorization.

I think though that's for sure super unclear right now.

With this, if all of the questions are the "same", then the bot should use "mode 1" for the survey, and group the questions together. If each of the questions are different", then the bot should use "mode 2". Either way, a better UX is probably needed.

An additional complication, is that discord seems to limit the length of questions (as noted above). This is why I put the questions under the description, and just had the survey response dialog show enter a response.

Totally open to improvements here, but hopefully this is helpful context.

@berkingurcan
Copy link
Collaborator Author

For now, in my branch, that you can see above from screenshots, the description can be much longer but in both case we cannot show the descripton to the user. They can only see and read before the click respond button.

So, I understand these modes like this and similar to my implementation:

  1. Mode 1: create command for single type surveys:
Screenshot 2024-05-27 at 20 45 43 Screenshot 2024-05-27 at 20 46 52 Screenshot 2024-05-27 at 20 47 32
  1. Mode 2: Create-multi-response command for up to max maxResponsesForMultiResponsePerUser questions. You can see the screenshots above.

In either case, I believe this version offers improvements over the previous one, as it maintains the functionality of the earlier model. Nevertheless, I don't have any idea for further enhancements to this situation.

@CristinaEche CristinaEche added enhancement New feature or request Survey Bot A bot to summarize and categorize surveys labels May 28, 2024
@CristinaEche
Copy link
Collaborator

tested last week and from survey admin point of view, now I can set specific questions for each header, but have to be careful to not exceed 45 char. in titles.

image

image

@berkingurcan is there anything else to do here? or we are just waiting for Evan's approval?

@berkingurcan
Copy link
Collaborator Author

Yes, if we are okay for this format, I will implement restriction to 45 char.

@CristinaEche
Copy link
Collaborator

Is there any other option? can we not apply the restriction of 45 char?

@berkingurcan
Copy link
Collaborator Author

Unfortunately not, it is a built in restriction and if we would like to see survey info in modal, we should apply.

@es92
Copy link
Collaborator

es92 commented Jun 6, 2024

To confirm, what would the different modes of submission be after this change? I can imagine having 2 modes:

  1. "Single Response" -- multiple questions allowed, but they get concatenated before they are sent to the LLM (each user gets one entry for all of their responses)
  2. "Multi Response" -- multiple questions allowed, but they do not get concatenated before they are sent to the LLM (each response is its own entry). Users can also answer the same question more than once (eg what are feature improvements you would like to see, list up to 5).

@berkingurcan
Copy link
Collaborator Author

To confirm, what would the different modes of submission be after this change? I can imagine having 2 modes:

1. "Single Response" -- multiple questions allowed, but they get concatenated before they are sent to the LLM (each user gets one entry for all of their responses)

2. "Multi Response" -- multiple questions allowed, but they do not get concatenated before they are sent to the LLM (each response is its own entry). Users can also answer the same question more than once (eg what are feature improvements you would like to see, list up to 5).

Yes single response is as you said.
For multi response, I did not understand what did you mean by concatenated. We can get and process them separately and user cannot answer the same question more than once.

@CristinaEche
Copy link
Collaborator

CristinaEche commented Jun 7, 2024

Hi @es92 and @berkingurcan , maybe I missed adding here the testing of multiresponse titles editing final results, here are some evidences:

Note: Title restricted from UI, so we can't exceed the 45 char. in the survey title

Setting up the questions

image

See the updated questions in the headers:

image

@CristinaEche
Copy link
Collaborator

CristinaEche commented Jun 10, 2024

FYI: detected that we cannot use "-" in a survey title. It let's us create the survey but if we want to edit the survey it show an error. usage of "-" will be restricted.

@es92
Copy link
Collaborator

es92 commented Jun 10, 2024

To confirm, what would the different modes of submission be after this change? I can imagine having 2 modes:

1. "Single Response" -- multiple questions allowed, but they get concatenated before they are sent to the LLM (each user gets one entry for all of their responses)

2. "Multi Response" -- multiple questions allowed, but they do not get concatenated before they are sent to the LLM (each response is its own entry). Users can also answer the same question more than once (eg what are feature improvements you would like to see, list up to 5).

Yes single response is as you said. For multi response, I did not understand what did you mean by concatenated. We can get and process them separately and user cannot answer the same question more than once.

It's about where comments get categories and counted -- In the current "multi-response" mode, each user can make comments that get categorized into different topics / subtopics. In the "single-response" mode, each user only gets one entry that goes into topics / subtopics. By concatenate I meant when there are multiple questions, concatenate the answers to those questions into a single response, that can be fed into the LLM to go under a single topic / subtopic.

Lmk if that helps a bit? On further reflection about talking to the talk to the city folks a few weeks ago though, I think there is probably a better strategy we could do here though in general. Instead of having multiple input types, we should probably have multiple visualization types. (Visualize as sentiment towards a vote, multiple pieces of feedback about a claim, answers to a question, etc). I don't think we need to do that now though, we should just proceed with existing improvements for the moment!

@berkingurcan
Copy link
Collaborator Author

Yes now the distinction is clear. But I will check and improve sending the concatenations as well as prompts :) to LLM.

@CristinaEche CristinaEche added PR 56 Pending product approval Ready to deploy but pending for approval labels Jun 10, 2024
@CristinaEche CristinaEche linked a pull request Jun 10, 2024 that will close this issue
@CristinaEche CristinaEche added approved-for-product and removed Pending product approval Ready to deploy but pending for approval labels Jun 10, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
approved-for-product enhancement New feature or request PR 56 Survey Bot A bot to summarize and categorize surveys tested!
Projects
None yet
Development

Successfully merging a pull request may close this issue.

3 participants