International Journal of Engineering Trends and Technology (IJETT) - Volume 35 Number 4 - May 2016 Evaluation of Trust in Facebook using GFTrust Algorithm Saket Kumar*, Shreya Shripathi*, Sanjukta Phukan*, Mrs. Ashwini K** *Department of ISE, SJB Institute of Technology, Kengeri, Bengaluru- 560060 **Asst Professor, Department of ISE, SJB Insititute of Technology, Kengeri, Bengaluru- 560060 Abstract— The immense growth Facebook as one of the most used Online Social Networks (OSNs) has made it to play a major role in connecting people together through OSNs. But the role of Trust in connecting people has become a major challenge and issue in present world. Since Nowdays the probability of malicious behaviour in OSNs has increased hence it is difficult to Trust anyone on Facebook . Hence, it is desirable to find out how much a person should trust one another in Facebook . The approach to answer this question is usually called trust inference through Trust Score. In this paper, we propose a new trust inference algorithm (Called GFTrust) based on the Genralized network flow concept. The algorithm, in addition to being simple, resolves some problems of Trust Evaluation in Facebook . The analysis of the algorithm demonstrates that GFTrust calculates the trust score which defines the degree of Trust for any given individual on Facebook. Keywords—generalized network flow, online social networks, path dependence, trust decay, trust evaluation I. INTRODUCTION In recent years, online social networks (OSNs) like Facebook, LinkedIn, and Twitter have become fundamental parts of our online lives, and their popularity is increasing day by day. Among the social networking sites, Facebook— with nearly 1.28 billion users worldwide as of the first quarter of 2014— has been one of the most successful ones in recent years when compared with other online social networks. Facebook nowadays is being used as a tool to connect with family, friends, colleagues, associates, known or unknown people based on the same interests or for different purposes such as professional works, advertising their brands and businesses, making profit or entertainment. Besides the revolution that Facebook has generated in online social networking, it has also introduced new threats to its users, because of the number of increasing users on Facebook as people share their ideas, views along with their images and personal informations (phone number, relationship ISSN: 2231-5381 status etc).Often users also use Facebook for chatting and texting one another for interacting between different users, that is when the Trust term comes into picture, as often we cannot trust to as to the user we are interacting with on the Facebook is authentic or not. Being a part of users’ daily lives, Facebook introduce new security concerns especially because of the potential exposure of huge amounts of personal information that the users share on Facebook. Some examples of threats to Facebook include creation of fake profiles, false facebook pages to get password and username, spamming etc . As a result, maintaining security and privacy of Facebook profiles by users has become a major challenge in recent years. “Trust in a person is a commitment to an action, based on a belief that the future actions of that person will lead to a good outcome (Goelback [2]).” In this paper, we mainly discuss the GFTrust Algorithm and how it can be implemented to calculate the trust score for any given individual in Facebook so as to gain either full Trust or no Trust on them. Additionally, we also show how users can protect themselves from any sort of unknown and untrusted users. Finally, based on the parameters of Facebook we use use our algorithm to show how a Trust Score is generated for a given individual so as to make sure the user can be trusted or not. The rest of the paper is organized as follows: Section II, gives the background, Section III discusses the concept of GFTrust algorithm and details of the algorithm. Section IV will discuss system architecture and module description. Section V gives proposed system. Section VI gives the module description. Lastly, Section VII will give us the conclusions & future work . http://www.ijettjournal.org Page 148 International Journal of Engineering Trends and Technology (IJETT) - Volume 35 Number 4 - May 2016 II. BACKGROUND In this section, we will first we first analyze the necessity of defining the two challenges that are path dependence and trust decay and then discuss about the trust evaluation. A. Path Dependence Path Dependence is a term that shows how a set of decisions a user faces for any given situations becomes limited by the decisions that the user had made in past, even though the user’s past circumstances are no longer relevant. Once a society becomes use to a given circumstance it is tough to change from that circumstance. For example if a person is used to of writing letters it is tough for him to send emails as he is used to and dependent upon writing letters as his path dependence makes it difficult for him to use the new technology. B. Trust Decay The term Trust Decay means the decay or loss in the amount of trust from one user to another there are several factors for the trust decay like distance from source, the tie strength between users and personality of users (some people may opt to trust others, while some others may opt to disrupt). Here we mainly consider the distance as a factor for the trust decay as more the distance can cause loss and leakage of information so the quality of service may also reduce because the distance will be more so it may lead to larger time taking and the data or information form one user to another user may reduce in the quality so there is a loss in quality of service when it comes to distance a s a factor and hence there is a decay in the trust from one user to other. B.1. Trust Evaluation The term Trust Evaluation means the evaluating of trust form one user to another user in Facebook, based on the parameters like comments, interactions and tags. Each user is evaluated based on the given parameters and a trust score is generated for the given user based on that score the user is categorised as being trusted or not. there is no flow send ; this situation shows no matter how good be the recommendation from intermediate nodes be s will never trust d . Finally , if fo = ∞, that means there is an infinite flow to be sent from s to d only with the condition that there should be enough paths from s to d, in this case s will trust d highly since it is a case of heavy flow. In this paper however we do not consider the case of fo=∞, we just consider the case of 0<f0<1, since it is more flexible and the node leakage will take the same role. That is, we can set a larger node leakage if fo is larger, and vice versa. B. Exploring the Leakage at nodes Here we actually set the node leakage associated to each nodes, because the trust decay is caused by nodes rather than the edges. Generally the node leakage may be caused by several factors like the type of users and distance from the users, while modeling however we need to take care of certain factors like, “amount of flow that leak in each node”. The GFTrust algorithm provides scheme to a framework for considering these factors for future. In our study we consider proportional leakage as often during the propagation from one user to other user the trust (flow) will shrink for every given intermediate node. There is also a possibility of fixed leakage where the trust (flow) is lost in a certain fixed amount. C. Assiging Capacity to Edges Here we actually assign the capacity to each and every edge based on its capacity, we use the Trust Value t(e) on the given edges to represent the capacity of the given edge. Using the value the Trust value cannot be overused and will be in the given bounded limit, also in case when there are more number of paths existing. Hence on basis of the above given analysis the GFTrust algorithm works as :1. 2. 3. Modelling trust decay with node leakage. Constructing generalized flow network. Calculating a near – optimal generalized flow. III. GFTRUST ALGORITHM DETAILS In this section, we introduce the details about the GFTrust algorithm, there are already several works done on trust evidence collection (e.g ., Golbeck [2], Massa and Avesani [27]), so we do not have to look for collecting information, we assume that there is a graph with the trust relationships and values of directly-connected neighbours are already available. A. Initial Flow Determination As discussed earlier about the problems hence considering those in mind we use the concept and assume the initial trust , we let the initial flow from s be f =1 as it is the most natural value of initial flow. However if the value of fo= 0 that means figure 1. Algorithm for GFNear Optimal 0 ISSN: 2231-5381 http://www.ijettjournal.org Page 149 International Journal of Engineering Trends and Technology (IJETT) - Volume 35 Number 4 - May 2016 IV. SYSTEM ARCHITECTURE Based on the above understandings we implement the GFTrust algorithm for Facebook and for that we use the architecture in (Fig. 2.) The system architecture helps us to understand better about the use and working of the GFTrust algorithm in the Facebook to calculate the Trust Score and to evaluate the Trust for the given person on facebook the range of Trust score is from 0 to 1 based on the category we place them. Generally the process starts with the request sender sending friend request to user and the user does the authorization of the request sender . c)Less Known: In this category we place the users to whom we don’t know actually but due to the number of mutual friends we accept them as our friends so we place such users under the less known category and the Trust Score assigned to them is 0.25 . B. Context Information The context information deals with the number of friends (mutual friends) request sender has with the user and the dynamicity of the request sender. The dynamicity involves the following based on which the user is classified as trusted or untrsuted :1. number of login counts of request sender . 2. number of posts, comments, uploads request sender. 3. active and inactive status request sender . The above parameters are used to decide how many times the request sender logs in to their profile, how many times they posts picture or comments and uploads their details on their profile and what was the time duration upto which the request sender was active from their profile. These parameters are taken in consideration to decide the authenticity of the given request sender on Facebook . C. Interaction The Interaction tells the number of times the request sender interacted with the user based on the frequency of the messages sent on messenger, and the number of times the request sender commented to the user and tagged him in comments to show the frequency of interaction between the request sender and the user . V. PROPOSED SYSTEM figure :- 2. System Architecture The entire architecture is divided into three mainfolds :1. Experience (Ex) . 2. Context Information (CI) . 3. Inetraction (I) . A. Experience The experience deals with the understanding about a person when we receive a friend request from a request sender on facebook , we often place the request sender into the below mentioned three categories :a) Close Friends: Close Friends are the ones to whom we give full trust as we know them very well hence we place them to category of close friends and that gives them a Trust Score of 1 . b) Friends: Friends are the ones to whom we are not so close but we know them so when we place a user in the friends category the trust score given to them is 0.75. ISSN: 2231-5381 We propose the GFTrust scheme,where we use a modified generalized network flow model for the purpose of trust evaluation and to generate Trust Score regardless the problem of path dependence and trust decay.There is a given threshold between the nodes u and v once that threshold is achieved, v is taken as trusted by u, but with a given limit on capacity. Here we consider 1 to be the measure for full trust, corresponding to a full trust. Our contributions are threefold :( 1) Our work is the first to address the two challenges of path dependence and trust decay simultaneously, in the domain of trust evaluation in Facebook. Also, we use a modified generalized flow model with leakage, which is a novel approach in trust evaluation. (2) As a flow-based model, GFTrust has the advantage of generality, while saving the normalization process. Moreover, it bears the properties of incentive compatibility and Sybiltolerance, and it coincides with the basic axioms that a trust model should meet. (3) We conduct extensive experiments on Facebook based on GFTrust, to actually calculate the Trust Score of any given user on Facebook, to prove the Authenticity of the user. http://www.ijettjournal.org Page 150 International Journal of Engineering Trends and Technology (IJETT) - Volume 35 Number 4 - May 2016 VI. MODULE DESCRIPTION In this section, we discuss step by step of how exactly the GFTrust will work on the Facebook and the Trust Score is calculated for any given individual , we also see how the security level is set for protection of users , into friends. If they don’t have any conversation means they are underlying the less known category. 4.Security Level: This security level is based on those categories. Each has a certain limit. Based on that security level, the comments will be posted on their wall. Some words will store as a keyword in database. If they used that word means it won’t display. If a post is longer used means it will stored in a trending areas. VII.CONCLUSION AND FUTURE WORK figure 3. Use case diagram The following are the steps that are actually used to complete the entire process in Facebook Trust Score generation :1. Preprocessing: In this preprocessing phase, all users want to register their details in our site. It is very important to store their data in secret. So users have some crucial and secret information, these will be stored in unreadable format to admin. Then login to their page. They have lot of functionalities over there. Facebook has increased disclosure of personal information by making more information available online. Despite all the proactive security monitoring technologies that are used by Facebook nowadays, there are still chances of people getting cheated by fake profiles and sometimes cyber attacks and information disclosure. In this paper, we have reviewed the most common problem on Facebook about gaining Trust over a particular individual, using the proposed GFTrust algorithm. We have also discussed some counter measures that can be used against the problem to cure untrusted profiles on Facebook in order to maintain privacy, integrity, and availability of data that is being shared by the users in their Facebook profiles. However, taking everything into consideration, users have a critical role in protecting their own information by adhering to a set of security guidelines. Indeed, users themselves are responsible for any content that they share in their profiles, ignoring the fact that malicious users may find a way to access such content and initiate undesirable activities against them. However there was leakage associated with the distance, and a type of proportion leakage. In future work, we will improve the design, and also try the approach to fix the value of leakage. REFERENCES [1] 2. Request: This project is mainly used to communication between users. For that, initially the user needs to find their friend in this site. This may be based on known member or unknown member. We can communicate with anybody through that request. This request will accept by opponent is based on their wish. They can accept or not. [2] [3] [4] [5] 3. True score: This will base on users communication. Communication will separate into three categories. If their communication is high, the trust score will high and they are categories into close friends. If their conversation is normal, they are categories ISSN: 2231-5381 [6] W. Jiang, J. Wu, F.Li, G. Wang, H. Zheng. Trust Evaluation in Online Social Networks using Generalized Netwrok Flow. IEEE 2015. J.Golbeck. Computing and spplying trust in web-based social networks. PhD thesis, University of Maryland,2005. J.Wu. Trust mechanisms and their applications in MANETs. Keynote speech in TrustCom’09, 2009. W. Jiang, G. Wang, and J.Wu. Generating trusted graphs for trust evaluation in online social netwroks. Future Generation computer systems,31:48-58,2014. K.D. Wayne. Generalized maximum flow algorithms.PhD thesis,Cornell University,1999. Statista - The Statistic Portal: Number of monthly active Facebook users worldwide. Online at http://www.statista.com/statistics/264810/numberof-monthly-active-facebook- users-worldwide/ (2014). Accessed April 2014. http://www.ijettjournal.org Page 151