By Anthony Vilarim Caliani
I would like to thank everyone that has supported me to make it possible! It was a very fun challenge and of course I've learned a lot!
Thank you very much guys!
Mr. Owlf is a Fake News detector concept and my M.B.A. conclusion project as well 😅
As you can see by the architecture Mr. Owlf has some components like DSS, MLS... And now we are going to understand them a little bit better.
The Data Stream Service component is responsible for data ingestion and some statistics calculation. In this project we are going to use two pre-classified files, but the main idea here is to look foward adding more data sources like data streaming, web scraping and etc... By the way we can describe the project workflow by the image below.
The Machine Learning Service component is responsible for data classification and score building. This component is not only a service, It is a library as well, and its workflow is described by the following image.
- Service: The "service" is represented by "AI Modeling" dashed box and it is part of the execution pipeline. It is important to say that this part will also generate our classifier and our vectorizer files to be used by "MLS" library later.
- Library: The "library" is represented by "Score" dashed box and it is used in Mr. Owlf API for example. It means that you can import this library in your own Python project, besides that, this library is responsible for calculating data score and two of its inputs are provided by "MLS" service (classifier and vectorizer file).
Both components, API and Front-End are being used to show that we can expose and consume our Fake News detector as a service.
Further information about these components can be found on API Docs and Front-End Docs.
- 🌎 mr-owlf-api: API
- 📦 mr-owlf-dss: Data Stream Service
- 🕹 mr-owlf-front: Web Application
- 🤖 mr-owlf-mls: Machine Learning Service
- 👨🔬 mr-owlf-poc: Proof of Concept
Let's see how we run this project...
# Ta dahh
./run.sh start
# When you finish...
./run.sh stop