Browsing by Author "Koushik, L."

Filter results by typing the first few letters
Now showing 1 - 2 of 2
  • No Thumbnail Available
    Item
    Detecting Suicide Risk Patterns using Hierarchical Attention Networks with Large Language Models
    (Association for Computational Linguistics (ACL), 2024) Koushik, L.; Vishruth, M.; Anand Kumar, M.A.
    Suicide has become a major public health and social concern in the world . This Paper looks into a method through use of LLMs (Large Language Model) to extract the likely reason for a person to attempt suicide, through analysis of their social media text posts detailing about the event, using this data we can extract the reason for the cause such mental state which can provide support for suicide prevention. This submission presents our approach for CLPsych Shared Task 2024. Our model uses Hierarchical Attention Networks (HAN) and Llama2 for finding supporting evidence about an individual’s suicide risk level. ©2024 Association for Computational Linguistics.
  • No Thumbnail Available
    Item
    Interns@LT-EDI: Detecting Signs of Depression from Social Media Text
    (Incoma Ltd, 2023) Koushik, L.; Anand Kumar, M.; LekshmiAmmal, R.L.
    In this paper we show our approach to solve the Shared Task on Detecting Signs of Depression from Social Media Text at LT-EDI@RANLP 2023. The given task is to classify the Reddit posts present in the dataset provided, into 3 levels of depression: ‘not depression’, ‘moderate’ and ‘severe’. We have attempted classifying the posts using two models. We have explored multiple models for this task. Three of which will be included in this paper. The first model uses sentiment labels automatically extracted using TextBlob with TF-IDF for feature extraction and support vector machines (SVMs) for classification. For the second model, we leverage a convolutional neural network architecture for feature extraction and classification. Lastly, the third model incorporates a Bi-LSTM architecture with GloVe embeddings for feature extraction and classification. All the above models also used SMOTE for oversampling the dataset. Through our experimentation, we aim to evaluate the effectiveness of these models in accurately identifying signs of depression in social media text. © 2023 LTEDI 2023 - 3rd Workshop on Language Technology for Equality, Diversity and Inclusion, associated with the 14th International Conference on Recent Advances in Natural Language Processing, RANLP 2023 - Proceedings. All rights reserved.