Schema Matching Based On Deep Learning Using LSTM Model

schema matching aims to find the correspondence between two or more schemas. It is needed in many systems, such as data warehouses or data integration systems, as well as in many database applications. That is why we need to use better and more efficient methods. Different types of approaches are taken to unify the data collected and merge it into a single schema by using deep learning methods. A neural network attempts to map, and highlight the information that is needed to be expected with a function and therefore, achieve a more acceptable predicted output. In this research, a new assignment between data and schema is proposed using a popular model: the recurrent neural network (RNN) long Short term memory model (LSTM). Through a set of simulations, we choose better hyperparameters to create a model that allows us to predict the correct schema for our training data, where we established the acceptable model based on accuracy. Thus, these results are conducted using free datasets from the Udemy course platform.