developed by the Recife Center for ced Studies and information on the topological characterization of the Oro- Systems(c.e.s.a.r)(http://Www.ceSa This social Aro SNS network is located in Brazil and it w ed with the The recommendation mechanism procedure utilizes the intention to facilitate the exchange of experiences of the graph topology of the SNs to filter and order a set of node student of the Center of Informatics and the software that have some important properties in relation to a given engineers at CESAR. The network is composed of a total of node V. The nodes of the resulting set are recommendations 634 nodes and 5076 edges of new edges that should be connected to the node Vi. The Some preprocessing was applied on the data, so that the creation of these new edges is used to improve the properties proposed algorithm could be implemented. The Oro-Aro of the node V, besides providing benefits to the entire allows the creation of a one-way relationship, i.e. a user can network in terms of friendship connection add another user to his list of contact without the need for The process of recommendation is divided in two steps: a ne other user to approve the link. This kind of network is filtering procedure followed by an ordering. Filtering is an similar to twitter, a microblog network. Therefore a filter important step, because it separates the nodes with higher was used to remove all one-way relationship, i.e. who did possibilities to be a recommendation, consequently reducing not have an inverse relationship. The procedure reduced by the total number of nodes to be processed in the network. 29% the number of edges and 8% in the number of nodes The ordering step considers some properties to put the most (isolated sub-networks were removed). This procedure was relevant nodes in the top of the resulting list. As expected, necessary to obtain a network with symmetric connections: the result depends on the user the recommendation is also most of the social networks just allow two-ways generated for. Therefore, we applied an adaptive solution relationshi using Genetic Algorithm Figure 1 shows a graphical representation of the Oro-Aro social network. Each node represents a user and the edges a two-way relationship between users Fig. 2. The graph shows the degree of the node versus the frequency ccurrence. This figure demonstrates the behavior of connected users in the Oro. Aro social network A. Filter The algorithm used to perform the filtering procedure employs the concept of the clustering coefficient, which is haracteristic in small world networks [1. Using the natural idea that "It's more probable that you know a friend of your presents a use havin theo ion of the oro Aro relationships betwee networks, eacn node friend than any other random person"as stated by Mitchell in [4], the filtering step is restricted to select nodes adjacent II. RECOMMENDATION MECHANISM to each node that is adjacent to the central node of the recommendation process. All nodes that can be reached with he proposed friend recommendation system is based on two hops are considered the structural properties of social networks. The topological Figure 3 shows an example of this network for a single haracteristics, the information an le metrics are derived user. The node ni is the central element of the global from the complex network theory. It is observed that these recommendation process. The nodes within circle"A"are types of networks are defined as being either small-world or directly connected to the node n The nodes within circle scale free [1],[2], [3]. This characteristic can be used to the "B"and outside circle"A"are selected by the filtering development of a reliable recommendation mechanism rocedure Figure 2 shows a plot of the degree of the node versus the e. This plot provides somedeveloped by the Recife Center for Advanced Studies and Systems (C.E.S.A.R) (http://www.cesar.org.br/ network is located in Brazil and it was developed with the intention to facilitate the exchange of experiences student of the Center of Informatics and the software engineers at CESAR. The network is composed 634 nodes and 5076 edges. Some preprocessing was applied on the proposed algorithm could be implemented. The Oro allows the creation of a one-way relationship, i.e. a user can add another user to his list of contact without the the other user to approve the link. This kind of network is similar to twitter, a microblog network. Therefore a filter was used to remove all one-way relationship, i.e. who did not have an inverse relationship. The procedure reduced by 29% the number of edges and 8% in the number of nodes (isolated sub-networks were removed). This procedure was necessary to obtain a network with symmetric also most of the social networks just relationship. Figure 1 shows a graphical representation of the Oro social network. Each node represents a user and the edges a two-way relationship between users. Fig. 1. Visual representation of the Oro-Aro social networks, each node represents a user having the relationships between users III. RECOMMENDATION MECHANISM The proposed friend recommendation system is based on the structural properties of social networks. The topological characteristics, the information and the metrics from the complex network theory. It is observed types of networks are defined as being either scale free [1], [2], [3]. This characteristic can be development of a reliable recommendation Figure 2 shows a plot of the degree of the node vers frequency of occurrence. This plot provides some developed by the Recife Center for Advanced Studies and http://www.cesar.org.br/). This social located in Brazil and it was developed with the facilitate the exchange of experiences of the nformatics and the software composed of a total of the data, so that the proposed algorithm could be implemented. The Oro-Aro way relationship, i.e. a user can without the need for to approve the link. This kind of network is . Therefore a filter way relationship, i.e. who did not have an inverse relationship. The procedure reduced by the number of nodes . This procedure was a network with symmetric connections; also most of the social networks just allow two-ways graphical representation of the Oro-Aro social network. Each node represents a user and the edges a Aro social networks, each node users show a link. MECHANISM The proposed friend recommendation system is based on s. The topological metrics are derived observed that these either small-world or can be used to the recommendation mechanism. Figure 2 shows a plot of the degree of the node versus the frequency of occurrence. This plot provides some information on the topological characterization of the Oro Aro SNS. The recommendation mechanism procedure utilizes the graph topology of the SNS to filter and order a set of that have some important properties in relation to a given node v . The nodes of the resulting set are of new edges that should be connec creation of these new edges is used to improve of the node v , besides providing network in terms of friendship connection. The process of recommendation is filtering procedure followed by an ordering. Filtering is an important step, because it separates the nodes with higher possibilities to be a recommendation, the total number of nodes to be processed The ordering step considers some properties to put the most relevant nodes in the top of the resulting list. the result depends on the user the generated for. Therefore, we applied using Genetic Algorithm. 0 20 40 60 80 100 120 140 1 3 5 7 9 11 13 15 17 19 21 23 frequency degree Fig. 2. The graph shows the degree of the node versus the frequency of occurrence. This figure demonstrates the behavior of connected users in the Oro-Aro social network. A. Filtering The algorithm used to perform the filtering procedure employs the concept of the clustering coefficient, which is characteristic in small world networks idea that “It’s more probable that you know a friend of your friend than any other random person” as stated by Mitchell in [4], the filtering step is restricted to to each node that is adjacent to the central node of the recommendation process. All nodes that can be reached with two hops are considered. Figure 3 shows an example of this network for a single user. The node ni is the central element of the recommendation process. The nodes directly connected to the node ni . The “B” and outside circle “A” are selected by the filtering procedure. information on the topological characterization of the Oro￾The recommendation mechanism procedure utilizes the graph topology of the SNS to filter and order a set of node rtant properties in relation to a given of the resulting set are recommendations connected to the node v . The creation of these new edges is used to improve the properties , besides providing benefits to the entire network in terms of friendship connection. e process of recommendation is divided in two steps: a filtering procedure followed by an ordering. Filtering is an important step, because it separates the nodes with higher to be a recommendation, consequently reducing nodes to be processed in the network. The ordering step considers some properties to put the most in the top of the resulting list. As expected, the user the recommendation is applied an adaptive solution 23 26 29 31 33 36 38 61 123 the degree of the node versus the frequency of the behavior of connected users in the The algorithm used to perform the filtering procedure the concept of the clustering coefficient, which is characteristic in small world networks [1]. Using the natural hat “It’s more probable that you know a friend of your friend than any other random person” as stated by Mitchell , the filtering step is restricted to select nodes adjacent node that is adjacent to the central node of the All nodes that can be reached with 3 shows an example of this network for a single is the central element of the global nodes within circle “A” are . The nodes within circle “B” and outside circle “A” are selected by the filtering 234