Study on Social Network Mental Disorder Detection Based Markov Model

With the explosive growth in popularity of social networking and messaging apps, online social networks (OSNs) have become a part of many people’s daily lives. There are many mental disorder encountered noticed of social network mental disorders (SNMDs), the basic parameter at which evaluate the mental level of user such as Cyber-Relationship Addiction, Information Overload, and Net Compulsion, have been recently noted. Symptoms of these mental disorders are usually observed day by day, resulting in delayed clinical intervention. In this work, the mining online social behaviour 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, now propose a machine learning framework, namely, Social Network Mental Disorder Detection (SNMDD), that exploits features extracted from social network data log file to accurately identify potential cases of SNMDs. We also exploit multi-source learning in SNMDD and propose a new SNMDD based Markov Model (SMM) to improve the accuracy. To increase the scalability of SMM, we further improve the efficiency with performance guarantee. Our framework is evaluated via 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.


INTRODUCTION
In the century we live in, which is called the Digital age, computer and internet have gained an absolutely central importance in human life, and social media had a prominent role in this picture. Besides easy, cheap and fast access to information through computers and the Internet, the expansion of the communication network is one of the factors that lead individuals to use social media. The statistics provided by Tayfun Acarer, who is the Chairman of the Information Technologies and Communication Authority in Turkey, reveal the recent level of social media addiction in Turkey. According to Acarer, monthly Internet usage has increased to 32-33 hours, which is the highest after Britain in Europe, and Turkey is among the countries that use the Internet most in the world, and it also takes the 4th range in social networking site usage hours and 6th range in the user number of social networking sites (www.yesilay.org.tr / Retrieved 18.03.2015). According to statistics, Facebook seems to be the most heavily used social networking site all over the world. Through Facebook, users are able to express themselves freely, establish new friendships and relationships, and follow and engage in dialogues with people or groups they are interested in. Social media has become an indispensable part of the communication age (Aygül, 2010: 95).
The mass appeal of social networks on the Internet could potentially be a cause for concern, particularly when attending to the gradually increasing amounts of time people spend online. On the Internet, people engage in a variety of activities some of which may be potentially to be addictive. Rather than becoming addicted to the medium per se, some users may develop an addiction to specific activities they carry out online. Specifically, Young [7] argues that there are five different types of internet addiction, namely computer addiction (i.e., computer game addiction), information overload (i.e., web surfing addiction), net compulsions (i.e., online gambling or online shopping addiction), cyber sexual addiction (i.e., online pornography or online sex addiction), and cyber-relationship addiction (i.e., an addiction to online relationships). SNS addiction appears to fall in the last category since the purpose and main motivation to use SNSs is to establish and maintain both on-and offline relationships (for a more detailed discussion of this please refer to the section on motivations for SNS usage). From a clinical psychologist's perspective, it may be plausible to speak specifically of 'Facebook Addiction Disorder' (or more generally 'SNS Addiction Disorder') because addiction criteria, such as neglect of personal life, mental preoccupation, escapism, mood modifying experiences, tolerance, and concealing the addictive behavior, appear to be present in some people who use SNSs excessively. The goal of above methodology is classifying nodes in graphs considering its position in the graph. Classic machine learning classification techniques assume that the label of each object is independent from others, however in a graph this assumption is not true and connected nodes will probably be in the same class. In other words, we cannot assume the label of each node as an independent entity from other nodes. For instance in a graph of friendship if a person is infected, her\his friends will have higher probability of infection. Another example can be in scientific paper author graphs, which represent the connection between nodes of the graph when they have at least one common author. In this data set the topic of connected papers are probably the same. In these 2 examples it is obvious that the label of data points is not independent of each other.
This idea is the motivation for us to explore new methods, which consider the dependencies between nodes in a graph. These methods are named "relational classification methods".
Our goal is to solve the collective classification problem when the labels of objects are related to each other.

V. RESULTS AND ANALYSIS
The loss function can be determined on basis given dataset and also different time frame. The time interval is 1 ms and result consider up to 7 ms.      When data size is 500 then value of memory utilization becomes 20 for SNMDD [1] and 15 for SNMD-MM (Proposed). Similarly when data size is 2500 ms then value of memory utilization 140 for SNMDD [1] and 135 for SNMD-MM (Proposed). Hence memory utilization of detection may decrease up to 3.5%. The accuracy can be determined on basis given dataset and also different number of iterations. The time interval is 5 and result consider up to 25.

VI. CONCLUSIONS & FUTURE SCOPE
An attempt to automatically identify potential online users with SNMD-MM. Now propose an SNMD-MM framework that explores various features from data logs of OSNs and a new tensor technique for deriving latent features from multiple OSNs for SNMD-MM detection. This work represents a collaborative effort between computer scientists and mental healthcare researchers to address emerging issues in SNMD-MM. On the basis of result and analysis, following observation is find out: 1. Loss content in detection process may decrease up to 24%. 2. Time complexity of detection process may decrease up to 0.34%. 3. Memory utilization of detection may decrease up to 3.5%. 4. Accuracy of detection may increase up to 1.13%.
As for the next step, we plan to study the features extracted from multimedia contents by techniques on NLP and computer vision. We also plan to further explore new issues from the perspective of a social network service provider, e.g., Facebook or Instagram, to improve the well-beings of OSN users without compromising the user engagement.