Enhancing the Accuracy of Emoji Conversion from Text using Novel Lagrangian Support Vector Machine Against Long Short-Term Memory Algorithm

This research study aims to employ a unique method that integrates Lagrangian Support Vector Machine and Long Short-Term Memory techniques to forecast emoji conversions from text on social media platforms. The primary focus of evaluation will be the prediction model's accuracy. This research study involves two distinct groups: the novel Lagrangian Support Vector Machine and Long Short-Term Memory models. The dataset utilized in this study focuses on predicting emojis using text and was obtained from Kaggle. The dataset is openly accessible and comprises 7,476 rows of textual data. In order to facilitate thorough evaluation, the dataset has been split into a training set, comprising 5,980 rows, and a testing set, consisting of 1,496 rows. The accuracy of the prediction models was calculated with a statistical power (G power) of 80%. The experimental study involved conducting twenty iterations using the aforementioned models. This study revealed that the Novel Lagrangian Support Vector Machine achieved an accuracy of 57.68% in predicting emojis from text, while the Long Short Term Memory model achieved an accuracy of 33.33%. Conducting a statistical analysis employing an Independent Sample T-Test with a p-value of 0.001 (significantly lower than the conventional significance level of 0.05), the study conclusively established that the Novel Lagrangian Support Vector Machine outperformed the Long Short Term Memory model by a significant margin in terms of performance. The findings suggest that the Novel Lagrangian Support Vector Machine approach is successful in accurately predicting emojis from text.