Ontology Design of Influential People Identification Using Centrality

Identifying influential people as a node in a graph theory commonly calculated by social network analysis. The social network data has the user as node and edge as relation forming a friend relation graph. This research is conducting different meaning of every nodes relation in the social network. Ontology was perfect match science to describe the social network data as conceptual and domain. Ontology gives essential relationship in a social network more than a current graph. Ontology proposed as a standard for knowledge representation for the semantic web by World Wide Web Consortium. The formal data representation use Resource Description Framework (RDF) and Web Ontology Language (OWL) which is strategic for Open Knowledge-Based website data. Ontology used in the semantic description for a relationship in the social network, it is open to developing semantic based relationship ontology by adding and modifying various and different relationship to have influential people as a conclusion. This research proposes a model using OWL and RDF for influential people identification in the social network. The study use degree centrality, between ness centrality, and closeness centrality measurement for data validation. As a conclusion, influential people identification in Facebook can use proposed Ontology model in the Group, Photos, Photo Tag, Friends, Events and Works data.


Introduction
The semantic web standard proposed by World Wide Web Consortium (W3C) as the next-generation web. Ontology chose as a knowledge-based semantic description for base formalisation. Ontology used to communicate between website on the Internet. Ontologies give more power on conceptual with domain and relation to describe the problems. Many research on software engineering use ontology as a modelling language for formalisation design. Especially multiple sites for communicating between designs in remote team member [1] to avoid miss-configuration of software and mediating mechanism for accessing heterogeneous data and information sources [2]. If machine learning and software framework can solve healthy problem [3] [4]. Data science can solve warehouse [5] and knowledge management problem [6]. Ontology is a solution for software engineering knowledge representation [1]. There was a classification of Ontology for Software Development Life Cycle phase [7]. Ontology is presenting more metadata schemas, offering a controlled vocabulary more than Unified Modelling Language (UML) [8]. Ontology can improve Social Network Analysis(SNA) measurement [9]and software engineering knowledge representation [10]. In the graph theory and network analysis, centrality used to determine the relative importance of a node within the graph [11]. Research topics in different aspects [12] using centrality analysis to get influential people. Beside for political issues [13] and terrorism information [14] on the social network [15], an online market business with higher values for three measurement degrees(degree, betweenness, closeness) obtain users who can target as productive customer for selling a product [16].
All the user-related graph on a social network mapped to semantic description model on ontology. Ontologies are a useful tool for modelling, sharing and reusing knowledge [10]. Ontologies are closely related to modern object-oriented software engineering with richer information model compared to UML and as a shared conceptualisation, naturally easy to develop in future work [8]. Centrality [17] use as graph theory with social network analysis [18] [19]. A person can be influential observed by how many friends list that he has, his ability to connect two networks of friends, or a person who knew a lot of others influential person. Degree, betweenness, and closeness centrality as popular centrality concept adopted in measuring graph. A graph consists a node and an edge. Node is defined person; edge defines a relation between nodes.
The objective of this research is modelling influential people from social network using OWL and RDF. Semantic description will use to modelling the concept of influential people. The relationships model used from a popular social network Facebook, which have ability providing user data in Application Programming Interface(API). The data gathered from Facebook using API [20]. Facebook allows third-party apps to take a user data from Facebook account who was interested in the application.

Methods
In the social network analysis, the data designed as structural characteristics of a social organisation. The Analysis need more information about the relation between nodes and edge [21]. Social network semantic make a computer understand the social network. In real social status, every person has a different relationship with different level and function. The Semantic describe more than attributes data and relation like characteristics of nodes and connections. Ontology is a philosophical concept to describe things. It is a modelling tool in artificial intelligence, which can describe conceptual models of information system in semantic and knowledge level [22] as a formal specification of a shared conceptualisation [23]. The definition relations in ontology illustrated as actor and relation attributes. The actor could be people, organisation or company and others [9]. Relation attributes describe various connection formed in social activities interact.
A Graph concept use for studying social phenomena. Actors and relationships are forming social networks data, the detailed explanation explains by the application of Wasserman and Faust (1994) [11] or Scott (1991) [24]. The most important thing in the analysis of social networks is vertices in the graph (Bavelas, 1984;Sabidusi, 1966;Freeman, 1979) [25]. The design was interpreting the interests into the graph. The formulation of centrality [10] obtained of the most influential people in a community. Centrality calculation divide into several calculation formulas.
Degree centrality is the most fundamental measure of centrality computation. Degree centrality has a formula essentially measures the amount of edge connection as a vertex of an object (equation 1). A nodes that could be a broker node or interface between node is a bridge between clusters. There will be no flow of information between two di without this bridge. Betweenness Centrality seeks each vertex connects directly to the vertex then measure it. dG(s, t) is the notation of the distance between vertex s to vertex t that the value of = 0 for all s ∈ V , and dG(s, t) = indicates the importance of a vertex in a graph. If from s ∈ V to t ∈ V , where the agreement from s to t where there is v ∈ Betweenness Centrality explains in equation 2.
s and t is a sum of any possibility of vertices in the graph. vertices. σ st (v) is all the number of shortest paths through v from s to t. paths from s to t.
Calculation of Closeness Centralit network. There is a node which is directly connected with the highest degree centrality value and connected with the highest points on the Betweenness Centrality.Closeness Centrality is the average distance (shortest path) between vertices connection. Closeness Centrality formula describes in Equation 3. CC (v) is Closeness centrality for vertex v. dG(v, t) is the distance between the vertex v to t.

Experiment and Result
The ontology defines actor and relation attributes into a property. Actor in this research identified as Social Network user account. One actor has many relation attributes. Facebook data have several relationship attributes. There is: HasGroup HasAttentEvent, HasTagged, HasLocation, HasHometown. Facebook user is an actor object as influential people. Two superclass is in a relationship as properties bridging actor with another object as classes. Classes is an object related to the user(figure 1).
Three Facebook classes create the domain on the social network (figure 2). The user is an account who use a social network. Area describes the city or town where the user comes from and lives in recently. Activities describe user activity in Facebook. Under activity class, there is a group where the user joins in. Photo tags where the user was in the photo with someone else. Events which are user come to an invitation. The school who is user getting a school the A nodes that could be a broker node or interface between two different clusters in a graph. This node is a bridge between clusters. There will be no flow of information between two di without this bridge. Betweenness Centrality seeks each vertex connects directly to the vertex then ) is the notation of the distance between vertex s to vertex t that the value of = dG(t, s) for s, t ∈ V . There are exist variations of the notation that importance of a vertex in a graph. If σ st = σ Calculation of Closeness Centrality measure the fastest and individuals who can monitor the network. There is a node which is directly connected with the highest degree centrality value and connected with the highest points on the Betweenness Centrality.Closeness Centrality is the average distance (shortest path) between vertices connection. Closeness Centrality formula describes in CC (v) is Closeness centrality for vertex v. dG(v, t) is the distance between the vertex v to The ontology defines actor and relation attributes into a property. Actor in this research identified as Social Network user account. One actor has many relation attributes. Facebook data have several relationship attributes. There is: HasGroup, HasPhotos, HasSchool, HasWork, HasFriend, HasAttentEvent, HasTagged, HasLocation, HasHometown. Facebook user is an actor object as influential people. Two superclass is in a relationship as properties bridging actor with another object s is an object related to the user(figure 1). Three Facebook classes create the domain on the social network (figure 2). The user is an account who use a social network. Area describes the city or town where the user comes from and lives in ities describe user activity in Facebook. Under activity class, there is a group where the user joins in. Photo tags where the user was in the photo with someone else. Events which are user come to an invitation. The school who is user getting a school there.
(1) erent clusters in a graph. This node is a bridge between clusters. There will be no flow of information between two different clusters without this bridge. Betweenness Centrality seeks each vertex connects directly to the vertex then ) is the notation of the distance between vertex s to vertex t that the value of dG(s, s) . There are exist variations of the notation that a notation of the number of shortest paths number of the shortest path between them. Based on the standard measurement of centrality, ) is Betweenness Centrality on is the number of shortest individuals who can monitor the network. There is a node which is directly connected with the highest degree centrality value and connected with the highest points on the Betweenness Centrality.Closeness Centrality is the average distance (shortest path) between vertices connection. Closeness Centrality formula describes in CC (v) is Closeness centrality for vertex v. dG(v, t) is the distance between the vertex v to  Photos is a list of user photos. Work is the occupation of the user. We can look clearly at Onto figure 3. Onto-graph is visualisation for domain and relation ontology, the connection between domain has many properties like in figure 4.
Super-properties in the social network ontology describing in five properties. HasActivity is a relation between users to the Activities superclasses, with HasUser as inverse properties. HasArea is a functional relation between user and Area Classes, with HasCitizen a HasFriend is a symmetric relation between users. There is a restriction in ontology, especially for the user. Property of come from has maximally related to one object  Photos is a list of user photos. Work is the occupation of the user. We can look clearly at Onto graph is visualisation for domain and relation ontology, the connection between domain has many properties like in figure 4. ties in the social network ontology describing in five properties. HasActivity is a relation between users to the Activities superclasses, with HasUser as inverse properties. HasArea is a functional relation between user and Area Classes, with HasCitizen as inverse function. Also, HasFriend is a symmetric relation between users. There is a restriction in ontology, especially for the user. Property of come from has maximally related to one object in HasSignificantTo and LiveIn.
Photos is a list of user photos. Work is the occupation of the user. We can look clearly at Onto-graph graph is visualisation for domain and relation ontology, the connection between domain ties in the social network ontology describing in five properties. HasActivity is a relation between users to the Activities superclasses, with HasUser as inverse properties. HasArea is a s inverse function. Also, HasFriend is a symmetric relation between users. There is a restriction in ontology, especially for the HasSignificantTo and LiveIn. Restriction rules added in user class. The user has relation to hometown maximal is one relation, same as location and significantTo. The semantic description of ontology for influential people identification generated from this ontology. The research use HasFriend and HasActivity or HasUser as properties and user and activities as classes. The Ontology is explaining in figure 5. In figure 6 the blue box is a class, a solid arrow showing subclasses. Dashed Arrow is showing properties of each class well connected each other. This ontology used by Social Network Analysis calculation. Measuring influential people using Centrality on this ontology to get a value of importance user as nodes in social activities and users graph.
Centrality calculation was implementing into Facebook user data. The data gathered by Facebook Application which have 120283 friend connection. The highest frequency of appearance of account in Degree Centrality data measurement, sorted by frequency of occurrence. The Result is account A get 3, account IMB get 3, account JRJS get 2 and account AW get 1.
For Each measurement of betweenness centrality in Social Network Data. The top three nodes selected by frequency sorting. The result is: account IMB get 2, account AL get 2, account TIH get 2, account AN get 2, account AM get 1.
The result from every closeness centrality measurement. The data sorting by frequency. Top three selected from population data. Account AL gets 3; account MI get 2, account ER get 2, account ES get 2, account KYT get 1.

Conclusion
The highest Centrality calculation using Equation 1, 2, and 3 have results for influential people identification using ontology model in Figure 5 and 6. Ontology model in Figure 5 and 6 in the Group, Photos, Photo Tag, Friends, Events, and Works data shows the name of a person, this identified as influential people. The centrality calculation from Location, Hometown data shows the name of City so cannot identify influential people. Education Shows the name of an institution so cannot identify influential people. As a result, influential people identification in Facebook can use ontology on Figure 6 in the Group, Photos, Photo Tag, Friends, Events and Works data.
Influential people identification can be more varied with different social network data. Centrality measurement for influential people identification can expand with other calculation method using fuzzy logic or another intelligence system. The model can combine with another technique besides centrality for evaluation purpose on influential people identification. Another aspect of individual influential people measurement can use statistical methods.