Wenyenchi.com
Brian Wendo
Nairobi, Nairobi County
- 0 Collaborators
Wenyenchi is a citizen engagement platform; an online baraza where citizens can engage directly with public and private service providers. It is a meeting point where experts from different parts of the world can assist community members to create homegrown solutions. ...learn more
Project status: Published/In Market
Intel Technologies
OpenVINO
Overview / Usage
The Accountability Project or 'wenyenchi' is an online platform where citizens can directly engage their leaders on matters development. The social impact objective of The Accountability Project is to give the Kenyan public a platform where they can directly engage public officials and the private sector leaders on service delivery, fiscal management, land, water and waste management, and response to community grievances. The goal is to create a community of problem solvers/reviewers across the country who use the platform to update the community on the development progress of their respective counties.
Methodology / Approach
Implementation of AI/ML components of The Accountability Project;
-
Sentiment Analysis - involves writing a python script that scrapes the reviews and stores in a text file which will be used for training the models. The reviews are a maximum of 1000 words roughly 3 and a half MS word pages. Hopefully we can get reviews using indigenous African languages to further enrich available datasets. The data can be used to identify top keywords and sentiments for an entity such as a company to understand their brand presence and perception.
-
Object detection - involves training the model to identify safety hazards (traffic violations, potholes), health hazards ( open sewers, garbage), environmental hazards ( dumping in rivers, deforestation, wildlife-human conflict, droughts, floods)
-
Image recognition - involves classifying roads into tarmac and mud/murram. We can use this to monitor the degrading of public facilities e.g. from a rating of 5 ( excellent condition) to 1 (poor condition)
-
Recommender system - involves suggesting similar reviews to users based on their reviewing history.
Technologies Used
Django, Bootstrap, Scikit-learn, Tensorflow, Sentry, SendGrid API, and Open Vino Toolkit, StreamLit ML