Declarative Experimentation in Information Retrieval using PyTerrier

07/28/2020
by   Craig Macdonald, et al.
0

The advent of deep machine learning platforms such as Tensorflow and Pytorch, developed in expressive high-level languages such as Python, have allowed more expressive representations of deep neural network architectures. We argue that such a powerful formalism is missing in information retrieval (IR), and propose a framework called PyTerrier that allows advanced retrieval pipelines to be expressed, and evaluated, in a declarative manner close to their conceptual design. Like the aforementioned frameworks that compile deep learning experiments into primitive GPU operations, our framework targets IR platforms as backends in order to execute and evaluate retrieval pipelines. Further, we can automatically optimise the retrieval pipelines to increase their efficiency to suite a particular IR platform backend. Our experiments, conducted on TREC Robust and ClueWeb09 test collections, demonstrate the efficiency benefits of these optimisations for retrieval pipelines involving both the Anserini and Terrier IR platforms.

READ FULL TEXT

Please sign up or login with your details

Forgot password? Click here to reset