The second ReNeuIR Workshop will be co-located with SIGIR in Taipei, Taiwan, in July 2023. The event aims to foster discussion and collaboration on holistic evaluation of methods in the age of neural information retrieval (NIR), noting that efficacy matters but so does the computational cost incurred to achieve it. In particular, the workshop promotes the following notions and encourages the community to raise and debate questions on the following themes:

• Justification: We believe it is important to justify the ever-growing model complexity through empirical analysis.

• Training and inference efficiency: We encourage the development of models that require less data or computational resources for training and fine-tuning, and that offer similarly fast inference. We also ask if there are meaningful simplifications of the existing training processes or model architectures that lead to comparable quality.

• Evaluation and reporting: We draw attention to the lessons learnt from past IR studies and encourage a multi-faceted evaluation of NIR models from quality to efficiency to environmental impact, and the design of reusable models, benchmarks, and standardized metrics.

More details on ReNeuIR 2023:

• Invited Speakers
• Organizers
• Program
• Call for Papers

For information on past ReNeuIR workshops, visit the links below:

• ReNeuIR 2022 at SIGIR in Madrid, Spain [Homepage, Report in SIGIR Forum]