Automated Score Calculation of Similar Words from Language Disabled Children using Hybrid Enhanced Latent Semantic Index

Children in modern days are born with disabilities which are highlighted as physical disability, language much more. Children ageing from 2-6 are mostly facing language or speech communication problems. Identifying the basic language for certain emotions or development of syntactic structure for specific sentences are most lacking in today's generations. Children with Speech Selective Language Impairment (SSLI) is the research problem focused in the current work where the proposed idea is handled with upcoming developments that are familiar for languages. Natural language processing is one of the advanced domains that works in language for automation and computation of the practiced structure. In existing works advanced Machine Learning (ML) applications with autoencoders, Recurrent Neural Network (RNN) techniques handled the ability of language and its lacks which shows the measures of low-level automation and Child’s reaction towards attempt of receiving words. Dataset from Kaggle popularly known as diagnose SLI in children uses a children language disability where there are samples used by Questionnaire and Answering (Q & A) sessions. To overcome the limitations in ML techniques in gathering the automation scores of similar words and language frames the proposed work Hybrid Enhanced Latent Semantic Index (HELSI) is introduced. Based on the structure, syntactic differences can be grouped and provided to the next level semantically which develop the lack of latent semantic analysis. Similarity score is compared based on implementation of metrics and accurate significant score is achieved with 98.4% of accuracy which is a reliable similarity score.