Automatic Target Generation for Electronic Data Interchange using Sequence-to-Sequence Models

Electronic Data Interchange (EDI) integration is crucial as it streamlines electronic data exchange between companies, enhancing efficiency and accuracy in business processes. This research aims to generate the corresponding target EDI format for given input data in a manner akin to the machine translation problem. The study explores various sequence-to-sequence models and encoder-decoder architectures to identify the most effective outcomes. The impact of these encoder-decoder and data input-output representation techniques on EDI integration is investigated using crowd-sourced datasets. The effectiveness of the proposed approach is demonstrated using different performance evaluation metrics. The findings revealed that using a Multi-Layer Seq2Seq model achieves an accuracy score of 90.21%, BLEU Score of 83.36% and ROUGE-L score of 88.5%, for enhancing EDI integration processes in businesses.