Emotion Prediction Using Semantic Analysis Neural Network

V. Arun, R. Vineeth and Ch. Prudhvi

Sentimental Analysis plays major role in today’s environment. Conclusion mining alludes to the utilization of common language handling, content investigation, computational semantics, and biometrics to deliberately recognize, separate, evaluate, and ponder full of feeling states and abstract data. Assessment investigation is broadly connected to voice the client materials, for example, audits and study reactions, on the web and online life for applications that go from promoting to client administration to film survey. This system proposes a novel way to deal with and recognize feelings like Good, Neutral or Bad in literary discussions or social media discussion utilizing a LSTM (Long Short Term Memory Networks) based Deep Learning model. Our methodology comprises of semi-robotized procedures to assemble preparing information for our model. This system misuse favorable circumstances of semantic and slant based embedding and propose an answer consolidating both. Our work is assessed on genuine world discussions and probably beats customary Machine Learning baselines just as other off-the-rack Deep Learning models. Based on vector based semantic organization, the projection to feeling space is executed as framework vector duplication among RE and other installing. At that point, the outcomes are forwarded to MCNN (Multi-Dimensional Convolutional Neural Network) to gain proficiency with an assumption classifier for microblog. This new model is named as EPUSAMCNN, short for Emotion-Prediction Using Semantic Analysis MCNN, which consistently incorporates feeling space projection dependent on emoticon into profound learning model MCNN to improve its capacity of catching feeling semantic. Our results look promising and indicate that neural nets are indeed capable of learning the emotion essayed by text.

Volume 11 | 04-Special Issue

Pages: 1184-1191