Detecting Veracity

Table of Contents

1. Introduction

2. Identification

3. Classification of Information

4. Architecture

5. Detection

6. Conclusion

7. References

Foreword

Mr. Hemant Kumar Saini is a Red hat Certified Engineer. He has obtained his M.Tech degree in Computer Science & Engineering in 2014.He has completed B.Tech from MLV Govt. Textile & Engineering College in 2011. He is the author of several articles published in reputed Journals and international Conferences.

Abbildung in dieser Leseprobe nicht enthalten

E-mail: hemantrhce@rediffmail.com

Abbildung in dieser Leseprobe nicht enthalten

Mr. NITESH CHOUHAN is Assistant Professor in MLV Government Textile & Engineering College, Bhilwara (Rajasthan). He has completed M.E. and B.E. in Computer Science & Engineering. He is having 9 years of academic experience. His research interests are Software Testing and Cyber Security.

1. Introduction

In the present era of the distributed system where almost the complete world has been engaged in social networking, how one can claim that he/she get the real authenticated content. Since the content on the internet is not verified especially the social media content where the people post mostly the doubtful information. The main difficult problem is the filtering of truth from such contents. In such a situation social media find the new challenge of establishing veracity (doubtable data). So a new system PHEME is going to establish for analyzing the content in social sites, blogs and socially related posts based on the language and determine the uncertainty or the doubts in the content. This system will help not only in medical information systems (where causes serious damages if the wrong information held) but also in digital journalism.

The system is under developing phase which claims to check the posts on social networks like Facebook and Twitter and then only permit to publish on Webpage. The project is running by International Group of Researches under the University of Sheffieldas part ofGATE, theUniversity of Warwick,and King’s College London,Saarland Universityin Germany andMODUL University Vienna.

2. Identification

Pheme is deployed with combination of the various techniques such as NLP, Data Mining, Web, Social Networking and Information visualization. With these techniques combine together it analyzes the realism and identify the information as shown in Fig.1. The various steps involved in it are discussed as follows:

- Firstly the lexical, semantic and syntactic information are collected from the data or the posts.
- Then the above collection is cross referenced with the Linked Open Data (LOD) through the expertise of Ontotext and various data sources using OWLIM platform.
- Next using the intelligent algorithms built in Sheffield‘s GATE text mining platform information is analyzed and detect it’s reality.
- Following that it will spin out results in a visual interactive mode with the different categories which are discussed in next section.

3. Classification of information

The fake information on the social network based on the semantics and language context can be classified into four categories:

- Speculation – unverified statements
- Controversial- disputed messages
- Misinformation – something untrue which spreads unwittingly
- Disinformation – This transmits intentionally with malicious interest.

These above categories are found by Ontotext which is also used by Pheme.

It is also advantaged by identifying malicious information with the analysis of history and semantics. Hence it search the sources like news outlines, individual journalist, automated bots etc that deny the information and plot evolvement of conversation on social media; network to assess whether it is diffuse with lies.

4. Architecture

The project is based on the WebLyzard and the complete architecture is divided into three layers -- front end, back end and the API as shown in Fig 2. The front end is the client browser where the data are fetched from the back end via websocket or HTTP requests. The complete interface is signaled with the bootstrap CSS framework where the JQuery creates and manages the websockets. With these it also make compatible for the mobile devices to report the correct viewtags.The back end separates the different components following the model view controller (MVC) design. When the pheme starts, template engine delivers HTML page with the compiled template. Each of these template get the log websocket server URL dynamically which access the database. And when the model updates the complete function is exported to SQL database. Here the new SQL directive enters into database will alert the controllers. For instance, the HTTP controller decides the view to be sent for the browser. Since MVC creates the list of events and whenever a new event occurs, the adapters which are always listening the communication layer convert the newly entered data into the model and register the object event using the RMI communication.But this architecture cannot be with-stand the continue logs so for such contiguous data a non blocking API is added which spawns its thread to buffering and sending the data even in connection lost. The API is designed in a jar file with the consideration of dependencies outside the java so that it can be easily deployed with the existing applications.

[...]