Conditional Questions on Event Types

[moilejter]

Is your proposal related to a problem?

We would like to have the ability to ask "follow-up questions" based on the answers to previous questions in the booking request. For example, we could have a mandatory question that asks "how did you hear about us?" with possible answers "web", "print", "social media", "personal reference" - and If they choose "web", we'd like to ask "source web site" (a text field), or if they choose "social media", we'd like to ask "social media site" (a drop down with entries for X/tweeter, Mastodon, Facebook, other).

Describe the solution you'd like

Add an option question type, which in addition to the existing question template, asks for the parent question identifier, and a value (or a list of values) that would make this question appear below the parent question.

Describe alternatives you've considered

We haven't come up with an alternative we liked.

Additional context

Did not find this already suggested in either "Issues" or "Pull Requests"

Requirement/Document

We haven't written one up.

[github-actions[bot]]

Thank you for opening your first issue, one of our team members will review it as soon as it possible. ❤️🎉

[hubsMIT1]

After reviewing, if this feature is needed, please assign it to me, and I will work on it.

[partik03]

can I take up this one?

[jimjoice2804]

could you give a little more detail please, thank you

[LalitDeore]

Hey, have you assigned this task to anyone? If not, then I would like to go with this task. Let me give you some details about how I can create this feature. Basically, you want to create the feature of a follow-up question based on the previous question. we can create context for each question; if the user clicks on any of the questions, we can show them a specific list of questions, and so on.

[moilejter]

Just so I know how this works - I am assuming that we have to wait for someone senior to approve this ticket, and then they would assign it to someone to work on? (I've gotten these notices - see above - about people who'd like to work on this, bu I figured it wouldn't be my place to ask anyone to work on this, right?)

[LalitDeore]

Ok, I am new to open source, and I thought you were someone who was managing this repository and raising this issue. That's why I asked you to work on this. Thanks for clearing.

[v-a14]

Hi, is there any update on this issue ?

[mkos11]

Proposed Solution:

Introduce a new question type, referred to as "Conditional Question," to the booking request form. This question type would have the following attributes:

1. Parent Question Identifier:

   • A field where the form creator can specify the identifier of the parent question. This identifier links the conditional question to its parent question.

2. Trigger Values:

   • A field or set of fields where the form creator can define the trigger values that, when selected in the parent question, activate the conditional question.

3. Question Template:

   • Similar to existing question types, this field allows the form creator to set the text or prompt for the conditional question.

4. Question Type:

   • Various question types (text field, drop-down, radio buttons, etc.) that can be selected for the conditional question.

Example Scenario:

1. Parent Question:

   • "How did you hear about us?"
     • Options: "web," "print," "social media," "personal reference"

2. Conditional Question 1:

   • Type: Conditional Question
   • Parent Question Identifier: [Identifier of "How did you hear about us?"]
   • Trigger Values: "web"
   • Question Template: "Source web site"
   • Question Type: Text field

3. Conditional Question 2:

   • Type: Conditional Question
   • Parent Question Identifier: [Identifier of "How did you hear about us?"]
   • Trigger Values: "social media"
   • Question Template: "Social media site"
   • Question Type: Drop-down with options for X/tweeter, Mastodon, Facebook, other

This way, when a user selects "web" in the "How did you hear about us?" question, the "Source web site" question appears dynamically. Similarly, if they choose "social media," the "Social media site" question becomes visible.

Implementation Considerations:

1. User Interface:

   • Ensure a user-friendly interface that allows form creators to easily set up conditional questions during the form creation process.

2. Validation:

   • Implement proper validation to handle cases where conditional questions may depend on multiple trigger values or complex conditions.

3. Flexibility:

   • Design the feature to be flexible, accommodating various question types for both parent and conditional questions.

4. Backend Logic:

   • Develop the backend logic to dynamically render and process conditional questions based on user responses to parent questions.

By introducing this "Conditional Question" feature, the booking request form gains enhanced flexibility, enabling more personalized and context-aware data collection.

[Adarshjain1338]

Hi, Is this issue is approved ? and if yes, Is it assigned to anyone ?

[ahsanfarooq210]

I would like to work on this issue. please assign it to me

[Maazchogle]

Proposed Solution:

                        AI-Driven Conditional Question Generation with Reinforcement Learning

1 Intent Recognition using NLP :

• Train a deep learning model such as BERT or RoBERTa to recognize user intents based on their responses
• Use pre-trained embeddings to capture semantic meaning
• Fine-tune the model on labeled data user responses and corresponding intents

2 Reinforcement Learning (RL) Agent :

• Create an RL agent that learns to dynamically generate conditional questions
• The RL agent interacts with the form, observes user responses, and receives rewards based on user satisfaction
• The agent's policy network generates questions and the reward signal guides its learning

3 State Representation :

• Represent the form state including user responses and context as a vector
• Include information about the parent question, previous answers, and any session context

4 Action Space :

• Define the action space for the RL agent:
• Generate a new question with placeholders
• Select an existing question from a predefined set
• Skip generating a question

5 Reward Function :

• Design a reward function that balances exploration and exploitation:
• Positive reward for relevant questions example matching intent
• Negative reward for irrelevant or redundant questions
• Encourage concise and context-aware questions

6 Training Process :

• Train the RL agent using Proximal Policy Optimization or other RL algorithms
• Simulate user interactions to collect training data
• Optimize the policy network to maximize cumulative rewards

7 Dynamic Question Generation:

• During form filling the RL agent generates questions based on the current state
• It considers the parent question, user responses, and learned context
• Questions are dynamically tailored to user intent

8 User Feedback Loop :

• Collect feedback from users:
• Did the generated question make sense?
• Was it relevant?
• Did it enhance the user experience?
• Use this feedback to update the RL agent's policy

Benefits of RL-Based Approach :

• Adaptability :

  • The RL agent learns from user interactions adapting to specific use cases
  • It improves over time as more data is collected

• Context Sensitivity :

  • The agent considers the entire conversation context not just trigger values
  • Conditional questions become contextually relevant

• Efficient Exploration :

  • RL balances exploration trying new questions and exploitation using learned policies
  • It avoids repetitive or unnecessary questions

Example :

Cal.com Event Registration Form :

Parent Question: How did you discover our upcoming event?

1 User Response: Via Email Newsletter

• Intent Recognition : The AI identifies the intent as "email channel"
• Conditional Question Generated :
• Question : Which section of the newsletter caught your attention?
• Question Type : Radio buttons with options like "Featured Events", "Spotlight Speakers" or "Workshops"

2 User Response: "Through Social Media Ads"

• Intent Recognition : The AI recognizes the intent as "social media advertising"
• Conditional Question Generated :
• Question : "Which platform did you see the ad on?"
• Question Type : Drop-down with options for platforms (e.g , Facebook, Instagram, LinkedIn)

3 User Response: "Word of Mouth"

• Intent Recognition : The AI detects the intent as "personal recommendation"
• Conditional Question Generated :
• Question : "Who recommended our event?"
• Question Type : Text field (for entering the referrer's name)

4 User Response: "Website Banner"

• Intent Recognition : The AI identifies the intent as "web banner"
• Conditional Question Generated :
• Question : "Which page on our website featured the event?"
• Question Type : Text field (for specifying the page URL)

5 User Response: "Event Listing on Community Forum"

• Intent Recognition : The AI understands the intent as "community forum"
• Conditional Question Generated :
• Question : "Which forum or community platform was it?"

• Question Type : Text field (for providing the platform name)

  AI-Driven Behavior:

  • The AI dynamically generates context-aware follow-up questions based on user intent
  • It considers the parent question and previous answers to tailor the form
  • If the user revises their response e.g , switches from "Email Newsletter" to "Social Media Ads" , the system adapts accordingly.

  Benefits of AI-Driven Approach:

  • Personalization :

  • Attendees receive relevant questions based on their unique journey

  • No generic or irrelevant queries

  • Efficiency :

  • The AI streamlines the form avoiding redundant questions

  • Users complete registration swiftly

  • User Satisfaction :

  • Attendees appreciate the thoughtful interaction

  • Conditional questions enhance engagement

Conclusion: Offering a smooth and customized user experience building an intelligent system that dynamically produces conditional questions by integrating reinforcement learning.

[tanyasnu09]

For this do we have to suggest through a writeup or a proper code should be posted here?

[keithwillcode]

@tanyasnu09 Feel free to submit a PR and we'll review!