Neural Information Retrieval
[14 June 2022] This was my first individual course project in the field of NLP. Fortunately, my model performed best in the Neural information Retrieval course offered in 2022 at the University of Copenhagen.
Background
This project was completed under the guidance of the Neural Information Retrieval (NIR) course at the University of Copenhagen. In this project we developed an information retrieval system for the news domain, including designing appropriate strategies for storing information, evaluating text retrieval models, and discussing experimental results. The code and report for this paper are available and its abstract is shown below:
This paper uses the NIR2022 dataset to evaluate various search algorithms: probabilistic and language models, divergence from randomness based models and neural retrieval models. In addition, query expansion techniques are used to improve effectiveness. The findings highlight the merit of embeddings, especially the direct use of them to compute similarities between queries and documents. The Cross-Encoder model using segments secured 1th position in the NIR course.