Welcome to the LimeSurvey Community Forum

Ask the community, share ideas, and connect with other LimeSurvey users!

Sharing a new AI tool: Convert survey documents into LimeSurvey LSS files (beta)

More
1 month 4 days ago #271491 by alexgu
Hi everyone, I’d like to share a tool we’ve been developing that might be useful for the community.It’s designed to automatically convert survey documents (e.g., Word or text format) into LSS files that can be directly imported into LimeSurvey. The tool is still in beta, but the current accuracy has already reached over 80%, which can save a lot of time compared to building surveys manually. The conversion is powered by an AI agent. You can try it out here: [url] aitoolsforsurvey.com [/url] We’re continuously improving the tool, and we’d love to hear your feedback, suggestions, or any issues you encounter. Community input will be very valuable to make it better. Thanks a lot!
The following user(s) said Thank You: holch, tpartner, eperezpena

Please Log in to join the conversation.

More
4 weeks 18 hours ago #272473 by holch
Looks interesting! Is there anything that needs to be considered to "make it easier" for the AI to recognize the question type or anything?

Does it work only in English or also in other languages? Couldn't find anything on the website on this topic. OK, never mind. Found it. Thank you for sharing!

Will try to test it soon and give some feedback.

Help us to help you!
  • Provide your LS version and where it is installed (own server, uni/employer, SaaS hosting, etc.).
  • Always provide a LSS file (not LSQ or LSG).
Note: I answer at this forum in my spare time, I'm not a LimeSurvey GmbH employee.
The following user(s) said Thank You: alexgu

Please Log in to join the conversation.

More
4 weeks 6 hours ago #272477 by alexgu
Thanks a lot for checking it out!

Right now the AI works best if the questionnaire has a clear layout — like numbered questions and options in bullet points or line breaks. That makes it easier for the system to guess the right question type.

And yes, it’s not just English — it can handle other languages too. Accuracy might vary a bit depending on the formatting, but we’re working on improving that.

Really appreciate you planning to give it a try — can’t wait to hear your feedback!

Please Log in to join the conversation.

More
4 weeks 6 hours ago #272479 by alexgu
Thanks a lot for checking it out!
Right now the AI works best if the questionnaire has a clear layout — like numbered questions and options in bullet points or line breaks. That makes it easier for the system to guess the right question type.
And yes, it’s not just English — it can handle other languages too. Accuracy might vary a bit depending on the formatting, but we’re working on improving that.
Really appreciate you planning to give it a try — can’t wait to hear your feedback!

Please Log in to join the conversation.

More
4 weeks 2 minutes ago #272487 by eperezpena
I got a error 500 while importing the LSS on

LimeSurvey Community Edition Version 6.15.10+250901

and the same on

LimeSurvey CE Version 3.17.5+190604

Please Log in to join the conversation.

More
3 weeks 6 days ago #272493 by alexgu
Thanks a lot for the feedback! The issue was caused by a bug — we’ve just fixed it.

Please try again and it should work now.

let me know if you still run into any trouble.

Please Log in to join the conversation.

More
3 weeks 5 days ago #272518 by Joffm
Hi,
I had no issues to import the lss.
Here a real small questionaire with my findings

Questionnaire was a Word document

Q1. How old are you? (open)
A question of "long text" was created. -> better a "short text" or a "numeric input"

Q2. How did you rate the food? (only if Q1>30)
- excellent
- very good
- good
- neither good no poor
- rather poor
- poor
- very poor
List(radio) with the correct condition

Q3. Which of these car brands do you like most?
- AUDI
- VW
- Mercedes
- FIAT
list(radio)

Q3a. Which of these car brands do you like?
- AUDI
- VW
- Mercedes
- FIAT
- none of them
list(radio)

Q3b. Which of these car brands do you like? (multiple)
- AUDI
- VW
- Mercedes
- FIAT
- none of them
multiple, so you have to mention this explicitly
No subquestions were entered

Q4.Please, tell us your sex?
A question of "long text" was created
Question was moved to the first group below Q1

Q5. At what time do you usually leave your house?
A question of "long text" was created

When I wanted to try something else today I got this error


Joffm

Volunteers are not paid.
Not because they are worthless, but because they are priceless

Please Log in to join the conversation.

More
3 weeks 5 days ago #272532 by eperezpena
Hi, overall it works very well and saves a lot of time in creating questionnaires, although some manual corrections are still needed.
  1. It generated a correct title even though it wasn’t specified in the TXT:
    Survey on the political and economic situation in Tucumán_20250911_194421
  2. It correctly identified the variables and the response categories.
  3. It created the groups exactly as it should.
    I marked them between questions with a semicolon:
GROUPS:
; Demographic Data
; Government Management
; Electoral Context
; Provincial Context and Economic Situation
; Previous Votes
  1. I specified the group relevance, but instead of assigning it to the group, it repeated it in every single question within the group:
    Code:
     PREVIOUS VOTES
    ; GROUP Relevance: (EDADABI.NAOK > 15 AND DONDEVO.NAOK < 999)
Here it had trouble with the “greater than” symbol, converting it into an HTML character:
(EDADABI.NAOK & gt ; 15) and (DONDEVO.NAOK & lt ; 999)

5) It defined age as Long free text, when Numerical input would have been better:
Code:
EDADBI
Could you please tell me your age?
; Question type: Numerical input
I tried to specify the variable type in this case.
  1. In questions like this, it detected the relevance I marked with a semicolon and applied it to the question:
    Code:
    ; ONLY FOR UNDECIDED (DECVOF.NAOK ==99)
    PROVOF
    And even if you haven’t decided yet, who are you more likely to vote for?
    1) Candidate A
    2) Candidate B
    3) Alberto D
    4) Candidate F
    96) Other electoral fronts
    97) Would vote blank or annul
    99) Don’t know / No answer
     
  2. Questions without response categories were defined as Long Free Text.
  3. It would be interesting to have a way to indicate relevances, question types, and group relevances more explicitly.

      

 

Please Log in to join the conversation.

More
3 weeks 4 days ago #272545 by alexgu
Thanks a lot for your very detailed feedback. it’s super helpful!
I’ll work on improving the question type recognition further based on your notes.

The error you ran into was due to a server issue on our side, and it has now been fixed.

Please feel free to try again anytime. hopefully it runs smoother for you this time.

Really appreciate your support and feedback!

Please Log in to join the conversation.

Moderators: tpartnerholch

Lime-years ahead

Online-surveys for every purse and purpose