Scalable Hashtag Recommender System


human2simpson

Hashtag Recommender System based on k-means clustering

A hashtag recommender system based on k-means, mini-batch fast k-means and a deep learning feature extraction phase.

The tool developed is a command line interface that takes as input an image and returns a list of hashtags recomandations.

The model built was trained on real data taken from Instagram, for the clustering phase Apache Spark was used for performance reasons.

For possible evaluation you can check the GitHub repository.

Skills used

  • Languages: Python, Scala
  • Concepts: Machine Learning, AlexNet, Kmeans, Instagram Hashtags
  • Tools: Spark, Flintrock, Pytorch, Hadoop
  • Stack: Amazon Web Services