Detecting depression using Hybrid models created using Google's BERT and Facebook's Fast Text Algorithms

T. Saravanan, T. Jhaideep, N. Hima Bindu

2022 2nd International Conference on Advance Computing and Innovative Technologies in Engineering (ICACITE) · 2022 · 39 citations · 9 references

Concepts

Abstract

Depression is the leading cause of worldwide disability and is also a leading cause of suicide. Depression has a significant impact on language use. We target the early diagnosis of depression in this study by applying several Machine Learning algorithms based on messages and posts on social media networks. CNN, SVM, Random-Forest, and Nave-Bayes are some of the Machine Learning algorithms. The model is trained using the BERT, Fast Text word embedding technique which was compared with the classification algorithm using Linguistic Metadata at the User Level.

References

9