stackoverflow-quality-rating

A machine learning enabled application to rate the question quality and suggest tags

Project Overview

Overview

Background

• Customer: Stackoverflow platform users
• Goal: To help the platform users in writing highquality questions with suggesting the right tags
• Pains:: Often poor question on stackoverflow creates a lot of problem for the user, many a times users gets suspended for repetitive behaviour for the same
• Gains:: Can help them in analyzing their question quality, and hence they can improve it. Other than that to improve their visibility, the suggested tags can be a great help too.

Value Proposition

• Product: The Product will roll out in two phases. * In the first phase we will see the product will able to rate the question quality. * In the second phase we will roll out the feature where users will get suggested tags for their question to improve their visibility.
• Alleviates: It will help the developers in reducing their stackoverflow reputation by overcoming poor questions asked.
• Advantages: They can now easily check their quality and hence no worry of negative rating. as well as the correct set of tags helps.

Objectives

• Create an automated pipeline to rate the question quality.
• Achieve > 85% for the classification problem on the defined metrics.
• Create another pipeline to suggest tags.

Solutions

• Core Features
  • ML service to classify the question ratings.
  • Continual learning on the new data.
• Secondary Features
  • ML service to suggest the tags for each question.
• Constraints
  • Maintain low latency
  • Suggest only tags from the tags disctribution to avoide irrelevant tags