Social Media Analytics: Enhanced sentiment analysis on social media videos containing child abuse content.
Name Ahmad Shaad Safwi
Student No. 17210439
Email Address Ahmad.safwi2@mail.dcu.ie
Programme MSc. In Computing, Data Analytics
Module Code CA640
Date of Submission Nov 24th 2017
Supervisor Gareth Jones
A report submitted to Dublin City University, School of Computing for module CA640 Professional and Research Practice, 2017/2018.
I understand that the University regards breaches of academic integrity and plagiarism as grave and serious. I have read and understood the DCU Academic Integrity and Plagiarism Policy. I accept the penalties that may be imposed should I engage in practice or practices that breach this policy. I have identified and included the source of all facts, ideas, opinions, viewpoints of others in the assignment references. Direct quotations, paraphrasing, discussion of ideas from books, journal articles, internet sources, module text, or any other source whatsoever are acknowledged and the sources cited are identified in the assignment references. I declare that this material, which I now submit for assessment, is entirely my own work and has not been taken from the work of others save and to the extent that such work has been cited and acknowledged within the text of my work. By signing this form or by submitting this material online I confirm that this assignment, or any part of it, has not been previously submitted by me or any other person for assessment on this or any other course of study. By signing this form or by submitting material for assessment online I confirm that I have read and understood DCU Academic Integrity and Plagiarism Policy (available at: http://www.dcu.ie/registry/examinations/index.shtml)
Name(s): Ahmad Shaad Safwi
Date: Nov 24th, 2017
Abstract
Social media is the fastest and one of the most popular ways of communication among all the sectors of the society these days, and as a result of which people use these social platforms to express their thoughts easily. Sometimes the stories are useful and sometimes they are violent and inappropriate and cause serious issues for society specially the growing up kids.
Child abuse is one of such cases which is increasing and getting provoked online everyday resulting long term effects including childhood trauma and effects on brain, damaging the life , personality and thinking of children . In this study, we present a supervised learning strategy for detecting child abuse videos present on the internet and marking them down before it gets uploaded causing a provoking chain among all frustrated parents, teachers, siblings and any other care takers.
Introduction
With the advancement of technology people can easily share thoughts, ideas, views and interests in public domain through social sites. It has brought a new age of fast communication enabling everybody to participate in public discussions. This gives everyone opportunity to put their thoughts in front of larger audience and also understand what is others opinion on the same issue/event/topic. Sharing has brought people together but there is another side of the coin too, which forces us to reconsider that are we getting close or going away by sharing views/content online? Though lack of rules and regulations has made social media an attractive stage for everyone to express their beliefs, opinions and emotions online, but everything goes wrong when the complex balance between freedom of expression and the defense of human dignity is disturbed. Indeed, there is a misuse of this platform in promoting offensive and hatred content, which deviates the regular users from their path of work and in some cases even, had some severe real life consequences. As the videos available at social media often reflects the real-world experiences of their users, there is an increased expectation that the norms of society will also apply in social media settings. There is an increasing demand for the social media platforms to empower users with the tools to report inappropriate content.
According to CDC (Centers for Disease Control and Prevention) child abuse is defined as “any act or series of acts of commission or omission by a parent or other caregiver (e.g. clergy ,coach ,teacher) that results in harm ,potential for harm ,or threat of harm to a child.
The World Health Organization (WHO) defines child abuse as “all forms of physical and /emotional ill-treatment, sexual abuse, neglect or negligent treatment or commercial or other exploitation ,resulting in actual or potential harm to the child’s health , survival ,development or dignity in the context of a relationship of responsibility ,trust or power.â€
Child abuse is of different types. It can be physical, sexual, emotional and child neglect. The occurrences of each type are regular and all of the types are vital .We cannot ignore any type of child abuse on going in current World. Social media have proved to be an alarming platform for such activities .There have been many campaigns running about it making people aware of the ill situations prevailing, but inspite of that social media is getting flooded everyday with new and new updates and uploads of videos having child abuse content e.g A child getting beated mercilessly by a school teacher or Child getting beaten up by housemaid. As far as awareness is concerned if users take this as an awakening call it might help but there are certain users who enjoy it and gets satisfaction to do it with their own kids or their relative’s kids /children basically such content is only provoking the feelings of all frustrated parents , mind takers etc. which leads to long lasting child abuse effects like severe fear throughout the life , hostility aggression and an entire disturbed social behavior. Emotional abuse can harm individual’s self -esteem ,socializing ability, lack of empathy which can lead to depression and anxiety. Social media is one of the major resources of promoting this act more and more ,though there are some websites which have manual intervention and screening of the content and when reported and identified are marked down but till that time it’s too late for that content to get published online.
To address this issue, I am proposing a solution to detect the videos having child abuse content beforehand and prevent them going online and flagging them down with appropriate error message by focusing on the content and the actions involved in the video with focusing on all new content worldwide.
Literature review:
Despite of abundance and large impact of online child abuse videos there have been a lack of published works regarding the issues. There are some very rare similar works done where artificial intelligence software was designed to sort any new or old child abuse media files available which can help police in catching the child abusers .The International research project Icop (Identifying and Catching Originators in P2P networks) founded by researchers at Lancaster University ,the German Research Center for Artificial Intelligence (DFKI) and University College Corp, Ireland combined automatic filename and media analysis techniques in an intelligent filtering module with the help of which it can distinguish between new criminal media and distinguish it from other media shared . Some related work regarding detecting CSA media in P2P networks were mainly focused on the ability to disrupt on-line child exploitation (e.g. (Joffres et al., 2011)), reliability issues regarding mutable identifiers such as identifying
IP addresses and GUIDs (e.g. (Liberatore et al., 2010b; Dai, 2010; Yang et al., 2013)) and also including the identification of key sharers
(e.g. (Westlake et al., 2011)).Some of the forensic tools was a result of this approach examples are CPS and RoundUp which were used not only for monitoring such activities but with the help of one can easily geolocate the act . Along with that the add on was having a centralized database assist law enforcement in their international struggle against online exploitation.Softwares such as PhotoDNA by Google and Microsoft enabled law enforcement to detect modified versions of known CSA media5 (Baluja, 2008; Microsoft, 2009)),but none of these tools offered support for identification of new or previously unknown child abuse media .
The authors of (Edwards and Rashid, 2012) demonstrated that how the collaborative filtering techniques that are typically used in recommender systems, can be successfully applied to identify new media in P2P networks of a certain category . Their method is based on the assumption that file-sharing traffic tends to cluster around interest, especially when it involves illegal content, such as CSA media. Hence, they were able to detect previously unknown examples from these categories without analysing their contents or filenames.
There is a recent work that was successful in detecting deception (Afroz et al., 2012),disguising behaviour (Rashid et al., 2013) and identifying paedophile activities Inches and Crestani, 2012) all of these models combined natural language analysis with machine learning techniques to get the desired output.
In recent years however there have been some work published on detecting videos.Majority of them focus on the utilization of skin based features (e.g. (Rowley et al., 2006) (Jones and Rehg, 2002), and (Fleck et al., 1996)) or bag-of-visual-word features (e.g. (Deselaers et al., 2008)).These techniques when applied to video data key frames from the video stream have to be drawn along with extracting image features like colour histograms and skin features (e.g. (Lee et al., 2006)) or skin area shapes (e.g. (Kim et al., 2008)). For video data, prior research also incorporated acoustic features, such as MFCCs (Zuo et al., 2008) or so-called audio words (Liu et al., 2011), which entails an analogue approach to the visual words technique, based on vector quantised audio features.
Research Plan :
As mentioned, in this area of work, there has been some research in textual analysis only. There is an obvious requirement to get into the roots of these acts of child abuse media presented online In this practicum, online video data would be collected which has content consisting of child abuse factors and types focusing on any type of child abuse incidents happening/happened . Post collecting sufficient data to construct and define the packages, video data will be processed. Result or expectation is to make a sytem/tool which is able to differentiate between videos consisting of child abuse content and the one having a happy flow path.
Conclusion:
There is no religion in this World that teaches and preach about child abuse .Every religion believes that children are the most precious gifts from the Almighty. There should be a strong hold on the line that differentiates child abuse and child treatment in society .We should use Social media for a purpose to spread peace and humanity more rather than using it for showing more and more child abuse content. People cross thresholds and land into child abusing while talking to kids and children whether they are grown up kids , newly born or any other kid age.As a result of which these acts disturb kids physically as well as psychologically throughout their life. There should be more and more campaigns ,ads and awareness progarmmes for all ages of society in order to aware them all what exactly child abuse is and what damage it is making to the children. Along with that strong actions should be taken on each and every one involved in child abuse act whether they are parents, teachers or any other care taker. Monitoring and getting into the roots of the accident in essential for child abuse rather than telecasting it on Global news or any other social platforms .In order to maintain world peace we should have domestic peace first ,if we all have attained that with love and care then there are chances to have world peace one day .