Multi-task learning on mental disorder detection, sentiment detection and emotion detection
dc.authorid | 0000-0001-5765-5735 | |
dc.authorid | 0000-0001-5765-5735 | en_US |
dc.contributor.advisor | Dehkharghani, Rahim | en_US |
dc.contributor.author | Armah, Courage | en_US |
dc.contributor.other | Işık Üniversitesi, Lisansüstü Eğitim Enstitüsü, Bilgisayar Mühendisliği Yüksek Lisans Programı | en_US |
dc.contributor.other | Işık University, School of Graduate Studies, Computer Science Engineering Master Program | en_US |
dc.date.accessioned | 2024-03-14T16:40:28Z | |
dc.date.available | 2024-03-14T16:40:28Z | |
dc.date.issued | 2024-02-12 | |
dc.department | Işık Üniversitesi, Lisansüstü Eğitim Enstitüsü, Bilgisayar Mühendisliği Yüksek Lisans Programı | en_US |
dc.department | Işık University, School of Graduate Studies, Computer Science Engineering Master Program | en_US |
dc.description | Text in English ; Abstract: English and Turkish | en_US |
dc.description | Includes bibliographical references (leaves 41-45) | en_US |
dc.description | x, 49 leaves | en_US |
dc.description.abstract | Suicidal behavior is a global cause of life-threatening injury and most of the time, death. Mental disorders such as depression, anxiety, and bipolar are prevalent among the youth in recent decades. Social media are popular platforms for individuals to post their thoughts and feelings on. Extracting people’s sentiments and feelings from such online platforms would help detect mental disorders of the users to treat them before it becomes too late. This thesis investigates the use of multi-task learning systems and single-task learning techniques to estimate behaviors and mental states for early diagnosis. I used data mined from Reddit, one of the popular social media platforms that provides anonymity. Anonymity increases the chances of individuals sharing what they truly feel in their real life. The obtained results by the proposed approaches open new doors to the understanding of how multi-task systems can increase the performance of text classification problems such as depression detection, emotion detection, and sentiment analysis, trained together in a multi-task learning network when compared to their training in isolation in a single-task learning network. We used the SWMH dataset, already labeled by 5 different depression labels (depression, anxiety, suicide, bipolar, and off my chest) and then added emotion and polarity labels to it and made it publicly available for researchers in the literature. The obtained results in this study are also comparable to other approaches in the field. | en_US |
dc.description.abstract | İntihar düşüncesi, dünya çapında, ömür boyu tehdit eden yaralanmaların ve çoğu zaman ölümün bir nedenidir. Depresyon, ankseyete bozukluğu ve bipolar gibi zihinsel bozukluklar, son yıllarda gençler arasında yaygındır. Sosyal medya, bireylerin duygu ve düşüncelerini paylaştıkları popüler platformlardır. Sosyal medya platformlardan insanların duygu ve düşüncelerinin çıkarılması, uzmanlar kullanıcıların zihinsel bozukluklarınını tespit edilmesine ve çok geç olmadan tedavi edilmesine yardımcı olacaktır. Bu tez, erken tanı için davranışları ve zihinsel durumları tahmin etmeye yönelik, çok görevli öğrenme sistemlerinin ve derin öğrenme tekniklerinin kullanımını araştırmaya çalışmaktadır. Anonimlik sağlayan popüler sosyal medya platformlarından biri olan Reddit'in metin verilerini kullandım. Anonimlik, bireylerin gerçek yaşamlarında hissettiklerini paylaşmasına artırır. Önerilen yaklaşımlarla elde edilen sonuçlar, çok görevli sistemlerin, izole eğitimlerine kıyasla birlikte eğitilen depresyon tespiti, duygu tespiti ve duygu analizi gibi metin sınıflandırma problemlerinin performansını nasıl artırabileceğinin anlaşılmasına yeni kapılar açmaktadır. Bu çalışmada elde edilen sonuçlar, alandaki diğer yaklaşımlarla da karşılaştırılabilir niteliktedir. | en_US |
dc.description.tableofcontents | Purpose of Study | en_US |
dc.description.tableofcontents | LITERATURE REVIEW | en_US |
dc.description.tableofcontents | Theoretical Background | en_US |
dc.description.tableofcontents | Related Works | en_US |
dc.description.tableofcontents | SUGGESTED APPROACH | en_US |
dc.description.tableofcontents | Data Pre-Processing | en_US |
dc.description.tableofcontents | General Methodology | en_US |
dc.description.tableofcontents | EXPERIMENTAL EVALUATION | en_US |
dc.description.tableofcontents | Dataset | en_US |
dc.description.tableofcontents | Used Programming Languages, Platforms and Tools | en_US |
dc.description.tableofcontents | Metrics for Evaluation | en_US |
dc.description.tableofcontents | Single Task with Attention Layer vs MTL with Attention Layer | en_US |
dc.description.tableofcontents | Single Task without Attention Layer vs MTL with no Attention Layer | en_US |
dc.description.tableofcontents | A subset of original dataset | en_US |
dc.description.tableofcontents | Illustrative table that shows basic elements to explain evaluation metrics | en_US |
dc.description.tableofcontents | Experimental results of f1-score from test dataset compared with other research in the field | en_US |
dc.description.tableofcontents | Sample predictions of custom single task and proposed MTL models | en_US |
dc.description.tableofcontents | Multitask learning frameworks according to relatedness | en_US |
dc.description.tableofcontents | The General view of the suggested Methodology | en_US |
dc.description.tableofcontents | Custom single-task model with attention layer | en_US |
dc.description.tableofcontents | Custom single-task model without attention layer | en_US |
dc.description.tableofcontents | The proposed MTL model | en_US |
dc.description.tableofcontents | Subset of Extended SWMH dataset after labelling by polarity and emotion labels | en_US |
dc.description.tableofcontents | Classification report of mental disorder detection using custom model that has an attention layer | en_US |
dc.description.tableofcontents | Classification report of sentiment detection using a custom model that has an attention layer | en_US |
dc.description.tableofcontents | Classification report of emotion detection task using a custom model that has an attention layer | en_US |
dc.description.tableofcontents | Classification report of MTL using a custom model with attention layer | en_US |
dc.description.tableofcontents | Classification report of mental disorder detection using the custom model without an attention layer | en_US |
dc.description.tableofcontents | Classification report of sentiment detection using the custom model without an attention layer | en_US |
dc.description.tableofcontents | Classification report of emotion detection using the custom model without an attention layer | en_US |
dc.description.tableofcontents | MTL classification report using the custom model without an attention layer | en_US |
dc.description.tableofcontents | Correlation matrix showing the relation between mental disorder labels, sentiment and emotion labels | en_US |
dc.description.tableofcontents | Confusion matrix showing mental disorder evaluations for both single task and MTL models respectively | en_US |
dc.description.tableofcontents | Confusion matrix showing sentiment detection evaluations for both single task and MTL models respectively | en_US |
dc.identifier.citation | Armah, C. (2024). Multi-task learning on mental disorder detection, sentiment detection and emotion detection. İstanbul: Işık Üniversitesi Lisansüstü Eğitim Enstitüsü. | en_US |
dc.identifier.uri | https://hdl.handle.net/11729/5919 | |
dc.institutionauthor | Armah, Courage | en_US |
dc.language.iso | en | en_US |
dc.publisher | Işık Üniversitesi | en_US |
dc.relation.publicationcategory | Tez | en_US |
dc.rights | info:eu-repo/semantics/openAccess | en_US |
dc.rights | Attribution-NonCommercial-NoDerivs 3.0 United States | * |
dc.rights.uri | http://creativecommons.org/licenses/by-nc-nd/3.0/us/ | * |
dc.subject | Natural language processing | en_US |
dc.subject | Multi-task learning | en_US |
dc.subject | Sentiment analysis | en_US |
dc.subject | Deep learning | en_US |
dc.subject | Emotion detection | en_US |
dc.subject | Doğal dil işleme | en_US |
dc.subject | Duygu analizi | en_US |
dc.subject | Duygu algılama | en_US |
dc.subject | Derin öğrenme | en_US |
dc.subject | Çok görevli öğrenme | en_US |
dc.title | Multi-task learning on mental disorder detection, sentiment detection and emotion detection | en_US |
dc.title.alternative | Zihinsel bozukluk tespiti, duygusallık(sentiment) tespiti ve duygu tespiti üzerinde çok görevli öğrenim | en_US |
dc.type | Master Thesis | en_US |
Dosyalar
Orijinal paket
1 - 1 / 1
Yükleniyor...
- İsim:
- Multi_task_learning_on_mental_disorder_detection_sentiment_detection_and_emotion_detection.pdf
- Boyut:
- 1.43 MB
- Biçim:
- Adobe Portable Document Format
- Açıklama:
- MasterThesis
Lisans paketi
1 - 1 / 1
Küçük Resim Yok
- İsim:
- license.txt
- Boyut:
- 1.44 KB
- Biçim:
- Item-specific license agreed upon to submission
- Açıklama: