LAUSR

Research has examined the use of user-generated data from online media as a means of identifying and diagnosing depression as a serious mental health issue that can have a significant impact on an individual's daily life. To achieve this, researchers have examined words in personal statements to identify depression. Besides aiding in diagnosing and treating depression, this research may also provide insight into its preva- lence within society. This paper introduces a Graph Attention Network (GAT) model for the classification of depression from online media. The model is based on masked self-attention layers, which assign different weights to each node in a neighbourhood without costly matrix operations. In addition, an emotion lexicon is extended by using hypernyms to improve the performance of the model. The results of the experiment demonstrate that the GAT model outperforms other architectures, achieving a ROC of 0.98. Furthermore, the embedding of the model is used to illustrate the contribution of the activated words to each symptom and to obtain qualitative agreement from psychiatrists. This technique is used to detect depressive symptoms in online forums with an improved detection rate. This technique uses previously learned embedding to illustrate the contribution of activated words to depressive symptoms in online forums. An improvement of significant magnitude was observed in the model's performance through the use of the soft lexicon extension method, resulting in a rise of the ROC from 0.88 to 0.98. The performance was also enhanced by an increase in the vocabulary and the adoption of a graph-based curriculum. The lexicon expansion method involved the generation of additional words with similar semantic attributes, utilizing similarity metrics to reinforce lexical features. The graph-based curriculum learning was utilized to handle more challenging training samples, allowing the model to develop increasing expertise in learning complex correlations between input data and output labels.