A Comprehensive Study on Social Network Mental Disorders Detection via Online Social Media Mining

ABSTRACT:

The explosive growth in popularity of social networking leads to the problematic usage. An increasing number of social network mental disorders (SNMDs), such as Cyber-Relationship Addiction, Information Overload, and Net Compulsion, have been recently noted. Symptoms of these mental disorders are usually observed passively today, resulting in delayed clinical intervention. In this paper, we argue that mining online social behavior provides an opportunity to actively identify SNMDs at an early stage. It is challenging to detect SNMDs because the mental status cannot be directly observed from online social activity logs. Our approach, new and innovative to the practice of SNMD detection, does not rely on self-revealing of those mental factors via questionnaires in Psychology. Instead, we propose a machine learning framework, namely, Social Network Mental Disorder Detection (SNMDD), that exploits features extracted from social network data to accurately identify potential cases of SNMDs. We also exploit multi-source learning in SNMDD and propose a new SNMD-based Tensor Model (STM) to improve the accuracy. To increase the scalability of STM, we further improve the efficiency with performance guarantee. Our framework is evaluated via a user study with 3126 online social network users. We conduct a feature analysis, and also apply SNMDD on large-scale datasets and analyze the characteristics of the three SNMD types. The results manifest that SNMDD is promising for identifying online social network users with potential SNMDs.

PROJECT OUTPUT VIDEO:

EXISTING SYSTEM:

• Kim et al. investigate the association of sleep quality and suicide attempt of Internet addicts. On the other hand, recent research in Psychology and Sociology reports a number of mental factors related to social network mental disorders.

• Chang et. al employ an NLP-based approach to collect and extract linguistic and content-based features from online social media to identify Borderline Personality Disorder and Bipolar Disorder patients.

• Saha et al. extract the topical and linguistic features from online social media for depression patients to analyze their patterns.

• Choudhury et al. analyze emotion and linguistic styles of social media data for Major Depressive Disorder (MDD). However, most previous research focuses on individual behaviors and their generated textual contents but do not carefully examine the structure of social networks and potential Psychological features.

DISADVANTAGES OF EXISTING SYSTEM:

• Although previous work in Psychology has identified several crucial mental factors related to SNMDs, they are mostly examined as standard diagnostic criteria in survey questionnaires.

• To automatically detect potential SNMD cases of OSN users, extracting these factors to assess users’ online mental states is very challenging. For example, the extent of loneliness and the effect of disinhibition of OSN users are not easily observable

• The developed schemes are not designed to handle the sparse data from multiple OSNs.

• The SNMD data from different OSNs may be incomplete due to the heterogeneity

PROPOSED SYSTEM:

• We argue that mining the social network data of individuals as a complementary alternative to the conventional psychological approaches provides an excellent opportunity to actively identify those cases at an early stage.

• In this paper, we develop a machine learning framework for detecting SNMDs, which we call Social Network Mental Disorder Detection (SNMDD).

• We propose an SNMD-based Tensor Model (STM) to deal with this multi-source learning problem in SNMDD.

• We propose an innovative approach, new to the current practice of SNMD detection, by mining data logs of OSN users as an early detection system.

• We develop a machine learning framework to detect SNMDs, called Social Network Mental Disorder Detection (SNMDD).We also design and analyzemany important features for identifying SNMDs from OSNs, such as disinhibition, parasociality, self-disclosure, etc. The proposed framework can be deployed to provide an early alert for potential patients.

ADVANTAGES OF PROPOSED SYSTEM:

Advantages of our approach are:

• The novel STM incorporates the SNMD characteristics into the tensor model according to Tucker decomposition; and

• The tensor factorization captures the structure, latent factors, and correlation of features to derive a full portrait of user behavior.

• We further exploit CANDECOMP/PARAFAC (CP) decomposition based STM and design a stochastic gradient descent algorithm, i.e., STM-CP-SGD, to address the efficiency and solution uniqueness issues in traditional Tucker decomposition.

• The convergence rate is significantly improved by the proposed second-order stochastic gradient descent algorithm, namely, STM-CP-2SGD.

• To further reduce the computation time, we design an approximation scheme of the second-order derivative, i.e., Hessian matrix, and provide a theoretical analysis.

MODULES:

• Data Collection
• Training
• Classification

MODULES DESCRIPTION:

Data Collection;

We first extract a number of social interaction features to capture the user behavior on social media.

Parasocial relationship (PR). The feature of parasocial relationship is represented as |aout|/|ain|, where |aout| and |ain| denote the number of actions a user takes to friends and the number of actions friends take to the user, respectively

Online and offline interaction ratio (ONOFF).As observed by mental health professionals, people who indulge themselves in OSNs tend to snub their friends in real life We extract the number of check-in logs with friends and the number of “going” events as an indicator of the number of offline activities to estimate the online (|aon|)/offline (|aoff |) interaction ratio.

Social capital (SC).Two types of friendship ties are usually involved in the theory of social capital i) Bond strengthening (strong-tie), which represents the use of OSNs to strengthen the relationships; and ii) Information seeking (weak-tie), which corresponds to the use of social media to find valuable information.

Social searching vs. browsing (SSB).The human appetitive system is in charge of the addictive behavior. It shows social searching (actively reading news feeds from friends’ walls) creates more pleasure than social browsing (passively reading personal news feeds).

Personal features: The duration, Self-disclosure based features,Temporal behavior features are extracted from thepersonal profiles of OSN users

Training:

The given SNMD features of N users extracted from M OSN sources, we construct a three-mode tensor T ∈ RN×D×M and then conduct Tucker decomposition, a renowned tensor decomposition technique, on T to extract a latent feature matrix U, which presents the latent features of each person summarized from all OSNs. We aim to feed these latent features for SNMD detection. Matrix U effectively estimates a deficit feature (e.g., a missing feature value unavailable due to privacy setting) of an OSN from the corresponding feature of other OSNs, together with the features of other users with similar behavior. Based on Tucker decomposition on T , we present a new SNMD-based Tensor Model (STM), which enables U to incorporate important characteristics of SNMDs, such as the correlation of the same SNMD sharing among close friends.

Classification:

Finally, equipped with the new tensor model, we conduct semi-supervised learning to classify each user by exploiting mini batch gradient descent algorithm. Mini-batch gradient descent is a variation of the gradient descent algorithm that splits the training dataset into small batches that are used to calculate model error and update model coefficients. Mini-batch gradient descent seeks to find a balance between the robustness of stochastic gradient descent and the efficiency of batch gradient descent. Finally predict what type of disorder.

SYSTEM REQUIREMENTS:

HARDWARE REQUIREMENTS:

• System             : Pentium Dual Core.

• Hard Disk        : 120 GB.

• Monitor           : 15’’ LED

• Input Devices  : Keyboard, Mouse

• Ram                : 1 GB

SOFTWARE REQUIREMENTS:

• Operating system : Windows 7.

• Coding Language : Python

• Database                : MYSQL

REFERENCE:

Hong-Han Shuai, Chih-Ya Shen, De-Nian Yang, Senior Member, IEEE, Yi-Feng Lan, Wang-Chien Lee, Philip S. Yu, Fellow, IEEE and Ming-Syan Chen, Fellow, IEEE, “A Comprehensive Study on Social Network Mental Disorders Detection via Online Social Media Mining”, IEEE Transactions on Knowledge and Data Engineering, 2018.