AmazonQA: A Review-Based Question Answering Task

Authors: Mansi Gupta, Nitish Kulkarni, Raghuveer Chanda, Anirudha Rayasam, Zachary C. Lipton

IJCAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate numerous models for answer generation and propose strong baselines, demonstrating the challenging nature of this new task.
Researcher Affiliation Academia Mansi Gupta , Nitish Kulkarni , Raghuveer Chanda , Anirudha Rayasam and Zachary C Lipton Carnegie Mellon University {mansig1, nitishkk, rchanda, arayasam, zlipton}@cs.cmu.edu
Pseudocode No The paper describes models and processes but does not include any pseudocode or algorithm blocks.
Open Source Code Yes To promote research in this direction, we publicly release2 the dataset and our implementations of all baselines. 2https://github.com/amazonqa/amazonqa
Open Datasets Yes We build upon the dataset of [Mc Auley and Yang, 2016]... We make a random 80-10-10 (train-development-test) split. To promote research in this direction, we publicly release2 the dataset and our implementations of all baselines. 2https://github.com/amazonqa/amazonqa
Dataset Splits Yes We make a random 80-10-10 (train-development-test) split. Each split uniformly consists of 85% of descriptive and 15% yes/no question types.
Hardware Specification No The paper does not provide any specific details about the hardware used for running the experiments (e.g., CPU, GPU models, or memory specifications).
Software Dependencies No The paper mentions using specific models like LSTM-based encoders/decoders and R-Net, and evaluation metrics like BLEU and ROUGE, but it does not specify version numbers for any software dependencies or libraries used.
Experiment Setup No The paper describes model architectures (e.g., LSTM-based encoder-decoder), feature engineering for the answerability classifier, and various baseline models. However, it does not provide specific hyperparameters like learning rates, batch sizes, number of epochs, or optimizer settings, which are crucial for reproducing the experiments.