Deep Learning and Natural Language Processing for Suicidal Ideation Using Instagram Posts

Darade, Vrushali Bhanudas (2023) Deep Learning and Natural Language Processing for Suicidal Ideation Using Instagram Posts. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (2MB) | Preview
Preview	PDF (Configuration manual) Download (1MB) | Preview

Abstract

Suicide is a global problem 1. Every year, more than 8 million people die due to suicide, which means one person commits suicide every 40 seconds 2. Identification of suicide ideation is the first step in saving a life. Traditionally, suicidal tendencies are identified using a questionnaire dataset, which the victims mostly provide. As data is provided by victims, its authenticity is dependent on various aspects, such as the purpose of collection, time of collection, who is collecting, whether is it anonymous and so on. Hence it will be difficult to exactly understand the emotions of the victims. The research aim of this project is to identify suicidal tendencies by analyzing Instagram images and captions, especially for the age group between 15 to 30. Analysing Instagram posts using Deep learning concepts such as image and text processing will help in understanding the emotion of the person who posted it, such as whether the person is happy, sad, depressed, or lonely. This information will be beneficial for consolers, psychologists, school and college administrators, and others who will provide appropriate treatment to the suffering person. To achieve the research goal, the Deep Learning algorithm VGG16 will be applied to image processing, as it has shown the best accuracy results in categorisation and Long-short term memory, along with CNN for text processing. The evaluation demonstrates high accuracy with VGG16, LSTM, and hybrid CNN+LSTM resulting in an excellent accuracy of 87% and 98% respectively. This demonstrates that images and captions posted on Instagram social media can be successfully processed for the identification of suicidal ideation. As the model produced excellent results, it can be extended to analyse various social media platforms such as Twitter, Facebook, Snapchat, and so on.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Staikopoulos, Athanasios UNSPECIFIED
Uncontrolled Keywords:	Deep learning; Natural Language processing; VGG16; LSTM; Instagram Post; Suicidal Ideation
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science P Language and Literature > P Philology. Linguistics > Computational linguistics. Natural language processing R Medicine > RA Public aspects of medicine > RA790 Mental Health Z Bibliography. Library Science. Information Resources > ZA Information resources > ZA4150 Computer Network Resources > The Internet > World Wide Web > Websites > Online social networks T Technology > TK Electrical engineering. Electronics. Nuclear engineering > Telecommunications > The Internet > World Wide Web > Websites > Online social networks
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Tamara Malone
Date Deposited:	17 May 2023 16:47
Last Modified:	17 May 2023 16:47
URI:	https://norma.ncirl.ie/id/eprint/6582

Actions (login required)

View Item