The historical reviews of users about items are often insufficient, thus making it difficult to ensure the accuracy of generated explanation text. To address this issue, we propose a novel model-ERRA with retrieval enhancement. With this additional information, we can generate more accurate and informative explanations.
Below are examples of how to run ERRA (with and without the retrieval enhancement). python -u main.py \ --data_path ./reviews.json \ --cuda 1 --checkpoint ./tripadvisorf/ \ --attention_mask >> 1 --use_retrieval >> 1
Our data reprocessing is based on CIKM20-NETE-Datasets.