Repository for master thesis Code: Term-based and Embedding-based Similarity Search in Large, Unstructured Data Sets

This repository provides code for creating Faiss-indices using different context-embedding models, as well as code for generating Elasticsearch indices.

Index Generation is conducted via the Index_generator-class. The Test-caller bash file is useful for generating indices automatically.

The code for comparing the models are found in experiments.py. There are also functions for generating test data sets.

The used splitter can be found at util/splitter.

The setup folder contains a list of package list in order to make a fresh Ubuntu installation ready for using this repository as easily as possible.