In this Github, there are two projects:
- Influencer Categorization : NLP in Python for influencers categorization on Instagram
- Next Basket Prediction : Deep learning in Python for Recommender Systems
- keyword extraction algorithm which includes: gensim_LDA, gensim_tf-idf, sklearn_tf-idf, Rake, Text_Rank
- keyword list is: Keyword = polling of keyword_extraction_algorithms + top n words +top m hashtags
- Keyword list maps to a predefined category list by word2vec
- This problem modeled as sequence prediction problem. Therefore, the RNN deep learning network was deployed
- To overcome sparsity, auto-encoder was used. The dataset and its feature vector was large
- The performance was measured by recall, precision and F1 score
Please read the paper: Temporal Behavioral Modeling for Recommendation A Deep Learning Approach_v2.pdf