短影音內容近年來風靡全球，成為各大社群媒體上的熱門娛樂形式。然而， 當代研究指出，短影音成癮會在生理和心理方面造成諸多負面影響，如注意力 下降和學習動機低落。此外，短影音成癮也與其他現象相關，例如缺乏現實生 活中的心理支持、家庭或學業壓力、以及社交恐懼等問題。然而，現今針對短 影音成癮的研究，多是在使用者出現生理、心理乃至生命負面後果時才被發 現。因此，本研究旨在構建以社群網路行為為基礎的短影音成癮資料集，並設 計短影音成癮的早期偵測架構。實際上，利用社群網路上的豐富使用者數位足 跡來解決心理健康問題是現今常用的方法。然而，過往研究多著重於憂鬱和自 殺傾向的偵測。在本研究中，我們首先引入大型語言模型來處理圖資料集中常 見的稀疏性問題和資料缺失問題;同時，我們將社群網路的行為數據分為不同 的模態，並設計了一個異質的社交網路結構作為短影音成癮偵測的主要依據。 我們使用自行建構的資料集對短影音成癮者進行一系列量化分析，並進行大量 實驗驗證利用社群數據與異質社群圖在短影音成癮偵測上的有效性。

Short-form video has become a globally popular form of entertainment in re- cent years, appearing on major social media platforms. However, contemporary research indicates that short video addiction can lead to numerous negative effects on both physical and psychological health, such as decreased attention span and reduced motivation to learn. Additionally, short-form video addiction (SFVA) has been linked to other issues such as a lack of psychological support in real life, family or academic pressure, and social anxiety. Currently, the detection of SFVA typically occurs only after users experience negative consequences. Therefore, the study aims to construct a short video addiction dataset based on social network behavior and design an early detection framework for short video addiction. In fact, leveraging the rich digital footprints left by users on social networks is a common method for addressing many psychological health detection issues. However, previous research has mostly focused on detecting depression and suicidal tendency. In this study, we first introduce large language models to address the common issues of sparsity and missing data in graph datasets. Meanwhile, we categorize social network behavior data into different modalities and design a heterogeneous social network structure as the primary basis for detecting short video addiction. We conduct a series of quantitative analyses on short video addicts using our self-constructed dataset, and perform extensive experiments to validate the effectiveness of our method EarlySD, using social data and heterogeneous social graphs in the detection of short video addiction.

Abstract (Chinese) I
Abstract II
Contents III
List of Figures VI
List of Tables VII
1 Introduction 1
1.1 BackgroundandMotivation ...................... 1
2 Related Works 6
2.1 MentalHealthDetectionwithSocialMediaData . . . . . . . . . . 6
2.2 GraphNeuralNetworkonSocialNetwork . . . . . . . . . . . . . . 8
2.3 Large Language Model as Enhancer on Graph-Related Task . . . . 9
3 Preliminary Analysis
11
3.1 RealDataCollection .......................... 11
3.2 FeatureExtraction ........................... 12
3.3 Quantitative Insight from demographic and social behavior feature . 13
4 Problem Formulation .......................... 18
5 Method Design 20
5.1 EarlySDArchitecture.......................... 20
5.2 LLM-enhancedStructuralAugmentation . . . . . . . . . . . . . . . 21 5.2.1 u-uEdgeAugmentation .................... 21 5.2.2 u-tEdgeAugmentation .................... 22
5.3 HeterogenousGraph-basedSFVAclassifier . . . . . . . . . . . . . . 23
5.3.1 EmbeddingInitialization.................... 23
5.3.2 Heterogeneous Graph Neural Network Classifier as SFVA EarlyDetector ......................... 24
6 Experiment 26
6.1 DataStatistics ............................. 26
6.2 Baselines................................. 27 6.2.1 Feature-basedclassifier..................... 27
6.2.2 State-of-the-art methods for mental health detection . . . . 28
6.3 ExperimentSetting........................... 29 6.3.1 SocialGraphRefinementModule ............... 29 6.3.2 HeterogenousGraphClassifier................. 29 6.3.3 EvaluationMetrics ....................... 30
6.4 PerformanceComparison........................ 30
6.5 AblationStudy ............................. 31
6.5.1 6.5.2 6.5.3
7 Conclusion
8 Appendix
ExpandedTopicSet ...................... 31
Similarity-based Edge Generation and Parameter Study ............................. 33 DifferentFeatureModality................... 34
36 37
8.1 FeatureDetails ............................. 37
Bibliography ............................. 40

[1] Julie Jargon. TikTok Brain Explained: Why Some Kids Seem Hooked on Social Video Feeds. The Wall Street Journal, 2022.
[2] Julie Jargon. This Was Supposed to be the Antidote for TikTok Brain. It’s Just as Bad. The Wall Street Journal, 2023.
[3] Jared Evitts. TikTok-addicted students delete app during exams. https: //www.bbc.com/news/uk-wales-62720657, 2022.
[4] Vanessa Yurkevich. Why experts worry TikTok could add to mental health crisis among US teens. https://edition.cnn.com/2023/01/11/ tech/tiktok-teen-mental-health/index.html, 2023.
[5] Mingming Li Fu Guo Chen, Yuhan and Xueshuang Wang. The effect of short-form video addiction on users’ attention. In Behaviour Information Technology, 2022.
[6] Jian Hong Ye, Yu-Tai Wu, Yu-Feng Wu, Mei-Yen Chen, and Jhen-Ni Ye. Effects of short video addiction on the motivation and well-being of chinese vocational college students. Frontiers in Public Health, 10, 2022.
[7] Honglei Mu, Qiaojie Jiang, Jiang Xu, and Sijing Chen. Drivers and con- sequences of short-form video (sfv) addiction amongst adolescents in china: Stress-coping theory perspective. International Journal of Environmental Re- search and Public Health, 19, 2022.
[8] Yinbo Liu, Xiaoli Ni, and Geng feng Niu. Perceived stress and short-form video application addiction: A moderated mediation model. Frontiers in Psychology, 12, 2021.
[9] Jiangfeng Yang, Yonghe Ti, and Yinghua Ye. Offline and online social support and short-form video addiction among chinese adolescents: The mediating role of emotion suppression and relatedness needs. Cyberpsychology, behavior and social networking, 2022.
[10] Teens and mental health: How girls really feel about social media.
https://www.commonsensemedia.org/sites/default/files/research/ report/how-girls-really-feel-about-social-media-researchreport_ web_final_2.pdf, 2022.
[11] Hamad Zogan, Imran Razzak, Shoaib Jameel, and Guandong Xu. Depression- net: Learning multi-modalities with user post summarization for depression detection on social media. SIGIR ’21, 2021.
[12] Ivan Mihov, Haiquan Chen, Xiao Qin, Wei-Shinn Ku, Da Yan, and Yuhong Liu. Mentalnet: Heterogeneous graph representation for early depression de- tection. ICDM’22, 2022.
[13] Jahandad Pirayesh, Haiquan Chen, Xiao Qin, Wei-Shinn Ku, and Da Yan. Mentalspot: Effective early screening for depression based on social contagion. CIKM ’21, 2021.
[14] Hong-Han Shuai, Chih-Ya Shen, De-Nian Yang, Yi-Feng Lan, Wang-Chien Lee, Philip S. Yu, and Ming-Syan Chen. Mining online social data for detect- ing social network mental disorders. In Proceedings of the 25th International Conference on World Wide Web, WWW ’16, page 275–285, 2016.
40
[15] Elton H. Matsushima Paulo Mann, Aline Paes. See and read: Detecting depression symptoms inhigher education students using multimodal social media data. (ICWSM’20, 2020.
[16] Ermal Toto, ML Tlachac, and Elke A. Rundensteiner. Audibert: A deep transfer learning multimodal classification framework for depression screening. CIKM ’21, 2021.
[17] Tao Gui, Liang Zhu, Qi Zhang, Minlong Peng, Xu Zhou, Keyu Ding, and Zhigang Chen. Cooperative multimodal approach to depression detection in twitter. AAAI’19, 2019.
[18] Max Welling Thomas N. Kipf. Semi-supervised classification with graph con- volutional networks. In International Conference on Learning Representa- tions, ICLR’17, 2017.
[19] Petar Veliˇckovi ́c, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Lio`, and Yoshua Bengio. Graph attention networks. In International Conference on Learning Representations, ICLR’18, 2018.
[20] William L. Hamilton, Rex Ying, and Jure Leskovec. Inductive representation learning on large graphs. In Proceedings of the 31st International Conference on Neural Information Processing Systems, NIPS’17, page 1025–1035, 2017.
[21] Keyulu Xu, Weihua Hu, Jure Leskovec, and Stefanie Jegelka. How powerful are graph neural networks? In International Conference on Learning Repre- sentations, ICLR’19, 2019.
[22] Chuxu Zhang, Dongjin Song, Chao Huang, Ananthram Swami, and Nitesh V. Chawla. Heterogeneous graph neural network. KDD ’19, 2019.

[23] Chao Huang, Xubin Ren, Jiabin Tang, Dawei Yin, and Nitesh Chawla. Large language models for graphs: Progresses and directions. In Companion Pro- ceedings of the ACM on Web Conference 2024, pages 1284–1287, 2024.
[24] Peisong Wang Jia Li Xiangguo Sun Hong Cheng Jeffrey Xu Yu Yuhan Li, Zhixun Li. A survey of graph meets large language model: Progress and future directions. In the 33rd International Joint Conference on Artificial Intelligence, 2024.
[25] Thomas Laurent Adam Perold Yann LeCun Bryan Hooi Xiaoxin He, Xavier Bresson. Harnessing explanations: Llm-to-lm interpreter for enhanced text-attributed graph representation learning. In International Conference on Learning Representations, ICLR’24, 2024.
[26] Zhikai Chen, Haitao Mao, Hang Li, Wei Jin, Hongzhi Wen, Xiaochi Wei, Shuaiqiang Wang, Dawei Yin, Wenqi Fan, Hui Liu, and Jiliang Tang. Ex- ploring the potential of large language models (llms)in learning on graphs. SIGKDD Explor. Newsl., 2024.
[27] Wei Wei, Xubin Ren, Jiabin Tang, Qinyong Wang, Lixin Su, Suqi Cheng, Junfeng Wang, Dawei Yin, and Chao Huang. Llmrec: Large language models with graph augmentation for recommendation. WSDM ’24, 2024.
[28] 陳淑惠,翁儷禎,蘇逸人,吳和懋,楊品鳳. 中文網路成癮量表之編製與心理計量 特性研究. 中華心理學刊, 45(3):279–294, 2003.
[29] 陳淑惠. 中文版社交互動焦慮量表 (social interaction anxiety scale- chinese version, sias-c).
[30] 李仁豪、鍾芯瑜. 中文版簡式「五大人格量表」BFI的發展. 測驗學刊, 2020.
[31] Bassam Hilal ALHarbi, Salama Aqeel Al-Mehsin, Jaafar Kamel Al-Rababaah, and Khaled Ahmed Abdel-Al Ibrahim. The Predictive Ability of Social Anx- iety within Internet Addiction among University Students. Journal of Edu- cation and e-Learning Research, 2021.
[32] Sung-Su Kim and Sung-Man Bae. Social anxiety and social networking service addiction proneness in university students: The mediating effects of experi- ential avoidance and interpersonal problems. Psychiatry Investigation, 19, 2022.
[33] Genevieve Dash, Wendy Slutske, Nicholas Martin, Dixie Statham, Arpana Agrawal, and Michael Lynskey. Big five personality traits and alcohol, nico- tine, cannabis, and gambling disorder comorbidity. Psychology of Addictive Behaviors, 33, 2019.
[34] Ahmet Rıfat Kayi ̧s, Seydi Ahmet Satici, Muhammet Yilmaz, Didem S ̧im ̧sek, Esra Ceyhan, and Fuad Bakiog ̆lu. Big five-personality trait and internet addiction: A meta-analytic review. Computers in Human Behavior, 63, 2016.