Standardization of Enumeration Data Based on Text Features Using Convolutional Neural Network with Long Short Term Memory

The rapid growth of Deep Learning (DL) techniques requires access to society due to their high impact on individual data, enabling the development of robust models. However, the high dimensionality of enumeration data that results from the complex dynamic environment lagged in adopting data-driven approaches. To address the aim is to advance stage-of-the-art by proposing a method to detect the enumeration data and evaluate it using DL-based techniques specifically Convolutional Neural Network (CNN) with Long Term Short-Memory (LSTM). The proposed approach standardization data, distributional data format representation with classification performance, implemented with DL model evaluated in terms of performance in several experiments. The proposed offers a comprehensive and unified approach for conducting enumeration data and outcome studies utilizing text features. The experimental results show that standardizing the data of text features significantly increases classification accuracy to 87.86%, precision to 76.89%, recall to 72.05%, F1-score to 72.83%, and AUC to 0.992 compared to the existing method of concatenating the Bidirectional Encoder Representations from Transformers (ConBERT), Transfer Learning (TL), Belief-Propagation – Recurrent Neural Network (BP-RNN) and Bidirectional Long Short-Term Memory (Bi-LSTM).

I. Introduction

Data standardization involves converting data formats to exchange data and ensuring the number of samples is consistent due to varying sizes. Consequently, data must be normalized to optimize the method for particular research [1]. In addition to standardizing reports, the software contains a semiology extrapolation function to automatically identify key semiology functions [2]. The inherent data schema helps overcome barriers to information sharing, promoting transparency and regulatory compliance, utilizing techniques such as Deep Learning (DL) [3]. Differences in data and techniques, as well as parameters, affect performance of trained data, including accuracy and precision [4]. To evaluate the effectiveness of proposed model, several experiments were conducted, which are essential for data as they allow different systems to exchange data consistently [5]. Data standardization ensures that companies have a complete and accurate picture of their data, allowing them to improve data sharing using DL techniques [6].