Exciting updates: Eduardo is now Assistant Profesor at MBZUAI (AI-focused university in Abu Dhabi), new timeline
Hello! It has been a while since the project campaign was successfull, and we owe you some updates. The delay has a very good reason however: soon after the project was funded, I (Eduardo) was invited for a faculty interview in Abu Dhabi at Mohamed bin Zayed University of Artificial Intelligence (MBZUAI), a graduate research university dedicated to advancing AI as a global force for humanity.
The great thing is: the interview was successful, I got the job, and moved to Abu Dhabi in August! It was been an unexpected whirlwind of activity, and during this time work on this project had to take a backseat. My faculty page is at https://mbzuai.ac.ae/study/faculty/eduardo-beltrame/
Now that things begin to settle down and once again I have time to focus, we have revised the project timelines, which in truth have been pushed back by about a year (moving halfway across the world and cultures takes some time!). The excellent news is that MBZUAI is just about the best place on earth to advance this project, due to the tremendous concentration of skilled students in researchers in the field of software engineering, machine learning, and natural language processing. MBZUAI has an entire department dedicated to natural language processing, and this means we can begin to think about data processing and analysis features beyond basic data collection features envisioned as part of v1 of the app (described at the end of this update).
For now, we will continue work on software development, and due to the school year timelines in Brazil, we expect that field deployment and trialing of the app will happen in March 2025. In the meantime, we will also be on the lookout for collaborators at MBZUAI to join this project and make it even better! Below some pictures and more details about MBZUAI.
Proposed scope of v1 of app prototype - data collection
The purpose of this project is to develop a piece of software that is easy to launch and enables researchers to do WhatsApp surveys. For version 1.0 we are keeping the scope of features as simple as possible. The proposed features and usage workflows are described below.
Deployment
Researcher provides Facebook WhatApp business API credentials (This includes the bot phone number)
Researcher providers server credentials (we will start with Google cloud only)
The software configures a WhatsApp bot automatically
Adding a new survey
Researcher uploads a csv spreadsheet file containing a list of survey questions.
The questions are turned into a WhatsApp survey (activated when a user messages the bot with the survey name)
End user experience
User sends a message to the bot with the name of the survey
Bot starts the survey and will sequentially ask the user the questions in the survey and records the answer.
For v1.0 we propose to only implement linear surveys. Each question may have "error handling" for invalid media types
Retrieving results
Researcher can download a csv file of survey results, with the text answers and the paths of media answers
Researcher can download folders with the media answers (audio, images, files)
Parsing csv files and converting them to Python survey logic
In order to allow for flexibility for future expansion, we propose to have a parser that will read the csv file and convert it into a Python file containing the survey logic (e.g. survey_logic.py).
This will allow us to in the future allow users to deploy more complex surveys with the csv file by changing the parser
It will also allow users who have Python programming experience to define their own survey_logic.py file
0 comments