This project consists of the files and Jupyter notebooks, used for the Master Thesis project, and the master thesis itself.

USING STRUCTURED INFORMATION FROM TAGS FOR BOOK RECOMMENDATIONS

by Melania Berbatova, MSc in Artificial Intelligence, Sofia University "St. Kliment Ohridski"

The project consists of the following notebooks, listed in the order they are used in the project:

1. Data preparation - Exploratory data analysis and selection of the ratings and book metadata in goodbooks-10k

2. Tags exploration and selection - Exploratory data analysis and selection of the tags in goodbooks-10k

3. RecSys - TensorRec - Design of a recommendation system in TensorRec

4. Tag test set for tag relations - Definition of relations between tags and creation of subset for testing algroithms for relationship learning

5. Learning relations between tags - Testing different methods for learning relashions between tags; Clustering with DBSCAN

6. Stacked embeddings - Experiments with stacked embeddings

7. Tags to hierarchies - Exploring clustering and co-occurences

8. Tags to genres relations - Merging the tags dataset to an ontology of genres

9. Tags to DBPedia Nodes - Mapping tags to DBPedia nodes