Idaka
Team Members
Table of Contents
Purpose
This project was created by the The Software Design Lab at the Universidad de los Andes and AI-Engineering at University of Passau. The major goal of Idaka is to facilitate access to good ML practices for SE, helping to avoid pitfalls and challenges that hinder the achievement of good performance of ML-enabled systems. Unlike current approaches, Idaka provide the basis of a model that retrieves practices based on the user’s query, so the users dont need to know or use predefined vocabulary to search for relevant practices.
Idaka as a web system is publicly available online on idaka.onrender.com
Video
Idaka functionalities overview
Three main features are proposed for the user to search according to their needs. For the main one, Idaka presents a query based search, it uses 2 possible methodologies. First, using a classic IR strategy in which documents relevant to a query are retrieved from a corpus based on a similarity metric. Second, using Alpaca as generative language model.
The second and third proposal uses the relations between stages, tasks, and practices as a basis. For the second one, Idaka offers a filter for the user to search practices by related tasks. The users can select one or more tasks they are interested in, and the platform will show the corresponding practices.
Finally, as a third way to search, the platform presents a visual way to follow the taxonomy of these mentioned relations. The users can select one of the stages, which takes them to their description and the list of related tasks. By clicking on one of the tasks, it shows the associated practices.
Idaka web system
Idaka as a web system is publicly available online on idaka.onrender.com
Also, you can download and compile Idaka locally. For this, please follow the complete instructions that are available in its Github repository. Those instructions will guide you through the processes of:
- Downloading the source code and data.
- Configuring a database with MobgoDB Atlas, as it is required by the tool.
- Configuring environment variables.
- Compiling and deploying Idaka.
Hosted on GitHub Pages - Theme by orderedlist