Optimizing semantic error detection through weighted federated machine learning: A comprehensive approach

Abstract

Recently, the improvement of network technology and the spread of digital documents have made the technology for automatically correcting English texts very important. In English language processing, finding and fixing mistakes in the meaning of words is a very interesting and important job. It is also important to fix wrong data in cleaning data. Usually, systems that find errors need the user to set up rules or statistical information. To build a good system for finding mistakes in meaning, it must be able to spot errors and odd details. Many things can make the meaning of a sentence unclear. Therefore, this study suggests using a system that finds semantic errors with the help of weighted federated machine learning (SED-WFML). This system also connects to the web ontology's classes and features that are important for the area of knowledge in natural language processing (NLP) text documents. This helps identify correct and incorrect sentences in the document, which can be used for many purposes like checking documents automatically, translating, and more. During its training and checking stages, the new model identified correct and incorrect sentences with an accuracy of 95.6% and 94.8%, respectively, which is better than earlier methods.