Citation Context

...etwork topology” has recently gained attractiveness, as several complex phenomena of scientific interest tend to manifest in those networks that are characterized by specific “topological properties” =-=[19, 47, 5, 20, 9]-=-. Thus, it is not surprising to find that a fundamental characteristic shared by recent approaches to network analysis is the central role played by a set of basic types of networks, identified by spe...

Citation Context

...ncy. 44. Connectedness. 45. Hierarchy. 46. Upper boundedness. 47. Average distance. 48-51. All pairs shortest path (min, max, avg, std). Formal definitions are available in Wasserman and Faust (1994) =-=[46]-=-. The metrics above are available in ORA [13]. 4.2 Design of Experiments Overall, the hypotheses we wish to test are: (1) stability, i.e., to what extent different sampling algorithms for the same pur...

Citation Context

...etwork topology” has recently gained attractiveness, as several complex phenomena of scientific interest tend to manifest in those networks that are characterized by specific “topological properties” =-=[19, 47, 5, 20, 9]-=-. Thus, it is not surprising to find that a fundamental characteristic shared by recent approaches to network analysis is the central role played by a set of basic types of networks, identified by spe...

Citation Context

...ave become a central theme for KDD [29, 15, 18, 24, 28] On the other hand, in the social and mathematical sciences, (social and complex) networks have been an object of research for a few decades now =-=[19, 37, 10, 21, 6, 7, 12, 11]-=-. Over the years, the communication across communities has increased, the major results of each discipline have been shared and assimilated by the others, and, occasionally, old ideas have resurfaced ...

Citation Context

... separability of the sampling algorithms surveyed or introduced in section 3.1. 2.1 Related Work The pure topology types we consider in the next section have been introduced separately over the years =-=[19, 38, 47, 5, 20, 9, 4, 39, 22]-=-. To the best of our knowledge neither exploratory nor compre1 The reference space used in this paper is defined by 47 metrics widely adopted in the social and physical sciences. We embed all sampled ...

Citation Context

...ave become a central theme for KDD [29, 15, 18, 24, 28] On the other hand, in the social and mathematical sciences, (social and complex) networks have been an object of research for a few decades now =-=[19, 37, 10, 21, 6, 7, 12, 11]-=-. Over the years, the communication across communities has increased, the major results of each discipline have been shared and assimilated by the others, and, occasionally, old ideas have resurfaced ...

Citation Context

... separability of the sampling algorithms surveyed or introduced in section 3.1. 2.1 Related Work The pure topology types we consider in the next section have been introduced separately over the years =-=[19, 38, 47, 5, 20, 9, 4, 39, 22]-=-. To the best of our knowledge neither exploratory nor compre1 The reference space used in this paper is defined by 47 metrics widely adopted in the social and physical sciences. We embed all sampled ...

Citation Context

...etwork topology” has recently gained attractiveness, as several complex phenomena of scientific interest tend to manifest in those networks that are characterized by specific “topological properties” =-=[19, 47, 5, 20, 9]-=-. Thus, it is not surprising to find that a fundamental characteristic shared by recent approaches to network analysis is the central role played by a set of basic types of networks, identified by spe...

Citation Context

... e.g., sub-samples of scale free networks are not scale free [43], at exploring the effectiveness of search strategies, e.g., greedy search finds short chains of acquaintances in small world networks =-=[33, 34, 32, 1]-=-, at developing models of information flow [41, 42] and information exchange [16], or at exploring the robustness of metrics for network analysis to variations in the topological properties [23, 8]. 3...

Citation Context

...In data mining and machine learning, the study of real world networks is essential for the development of sound theoretical models, which ultimately lead to unbiased inferences and robust predictions =-=[35, 40, 44, 25, 26, 30, 36, 2, 45, 31, 3]-=-. Analyses of real networks rely upon two crucial tasks: (1) that of “generating,” or “sampling,” networks that display realistic properties of interest, and (2) that of “determining” which pure topol...

Citation Context

...In data mining and machine learning, the study of real world networks is essential for the development of sound theoretical models, which ultimately lead to unbiased inferences and robust predictions =-=[35, 40, 44, 25, 26, 30, 36, 2, 45, 31, 3]-=-. Analyses of real networks rely upon two crucial tasks: (1) that of “generating,” or “sampling,” networks that display realistic properties of interest, and (2) that of “determining” which pure topol...

Citation Context

...In data mining and machine learning, the study of real world networks is essential for the development of sound theoretical models, which ultimately lead to unbiased inferences and robust predictions =-=[35, 40, 44, 25, 26, 30, 36, 2, 45, 31, 3]-=-. Analyses of real networks rely upon two crucial tasks: (1) that of “generating,” or “sampling,” networks that display realistic properties of interest, and (2) that of “determining” which pure topol...

Citation Context

... e.g., sub-samples of scale free networks are not scale free [43], at exploring the effectiveness of search strategies, e.g., greedy search finds short chains of acquaintances in small world networks =-=[33, 34, 32, 1]-=-, at developing models of information flow [41, 42] and information exchange [16], or at exploring the robustness of metrics for network analysis to variations in the topological properties [23, 8]. 3...

Citation Context

...ave become a central theme for KDD [29, 15, 18, 24, 28] On the other hand, in the social and mathematical sciences, (social and complex) networks have been an object of research for a few decades now =-=[19, 37, 10, 21, 6, 7, 12, 11]-=-. Over the years, the communication across communities has increased, the major results of each discipline have been shared and assimilated by the others, and, occasionally, old ideas have resurfaced ...

Citation Context

...nic texts have made available huge amount of “networked data,” to the data mining community at large, to the effect that models and algorithms for network analysis have become a central theme for KDD =-=[29, 15, 18, 24, 28]-=- On the other hand, in the social and mathematical sciences, (social and complex) networks have been an object of research for a few decades now [19, 37, 10, 21, 6, 7, 12, 11]. Over the years, the com...

Citation Context

... e.g., sub-samples of scale free networks are not scale free [43], at exploring the effectiveness of search strategies, e.g., greedy search finds short chains of acquaintances in small world networks =-=[33, 34, 32, 1]-=-, at developing models of information flow [41, 42] and information exchange [16], or at exploring the robustness of metrics for network analysis to variations in the topological properties [23, 8]. 3...

Citation Context

... e.g., sub-samples of scale free networks are not scale free [43], at exploring the effectiveness of search strategies, e.g., greedy search finds short chains of acquaintances in small world networks =-=[33, 34, 32, 1]-=-, at developing models of information flow [41, 42] and information exchange [16], or at exploring the robustness of metrics for network analysis to variations in the topological properties [23, 8]. 3...

Citation Context

...etwork topology” has recently gained attractiveness, as several complex phenomena of scientific interest tend to manifest in those networks that are characterized by specific “topological properties” =-=[19, 47, 5, 20, 9]-=-. Thus, it is not surprising to find that a fundamental characteristic shared by recent approaches to network analysis is the central role played by a set of basic types of networks, identified by spe...

Citation Context

...In data mining and machine learning, the study of real world networks is essential for the development of sound theoretical models, which ultimately lead to unbiased inferences and robust predictions =-=[35, 40, 44, 25, 26, 30, 36, 2, 45, 31, 3]-=-. Analyses of real networks rely upon two crucial tasks: (1) that of “generating,” or “sampling,” networks that display realistic properties of interest, and (2) that of “determining” which pure topol...

Citation Context

...oviding intuitions and mathematical theory that describe what happens to topological properties when only partial information is available, e.g., sub-samples of scale free networks are not scale free =-=[43]-=-, at exploring the effectiveness of search strategies, e.g., greedy search finds short chains of acquaintances in small world networks [33, 34, 32, 1], at developing models of information flow [41, 42...

Citation Context

...In data mining and machine learning, the study of real world networks is essential for the development of sound theoretical models, which ultimately lead to unbiased inferences and robust predictions =-=[35, 40, 44, 25, 26, 30, 36, 2, 45, 31, 3]-=-. Analyses of real networks rely upon two crucial tasks: (1) that of “generating,” or “sampling,” networks that display realistic properties of interest, and (2) that of “determining” which pure topol...

Citation Context

...ave become a central theme for KDD [29, 15, 18, 24, 28] On the other hand, in the social and mathematical sciences, (social and complex) networks have been an object of research for a few decades now =-=[19, 37, 10, 21, 6, 7, 12, 11]-=-. Over the years, the communication across communities has increased, the major results of each discipline have been shared and assimilated by the others, and, occasionally, old ideas have resurfaced ...

Citation Context

...ifiers, such as naïve Bayes (based on Multinomial and Poisson distributions), logistic regression, maximum entropy, SVM (with a linear kernel), voted perceptron, decision trees and k-nearest neighbor =-=[14]-=-. The results below correspond to the Poisosn flavor of naïve Bayes classifier that turned out to be more accurate in predicting the topology type of a given network [3]. In order to estimate the pred...

Citation Context

...ave become a central theme for KDD [29, 15, 18, 24, 28] On the other hand, in the social and mathematical sciences, (social and complex) networks have been an object of research for a few decades now =-=[19, 37, 10, 21, 6, 7, 12, 11]-=-. Over the years, the communication across communities has increased, the major results of each discipline have been shared and assimilated by the others, and, occasionally, old ideas have resurfaced ...

Citation Context

...ave become a central theme for KDD [29, 15, 18, 24, 28] On the other hand, in the social and mathematical sciences, (social and complex) networks have been an object of research for a few decades now =-=[19, 37, 10, 21, 6, 7, 12, 11]-=-. Over the years, the communication across communities has increased, the major results of each discipline have been shared and assimilated by the others, and, occasionally, old ideas have resurfaced ...

Citation Context

...34, 32, 1], at developing models of information flow [41, 42] and information exchange [16], or at exploring the robustness of metrics for network analysis to variations in the topological properties =-=[23, 8]-=-. 3 Pure Topology Types We begin with a heuristic description of the pure topology types. The sampling algorithms in the next section provide these intuitive notions with a precise meaning. Without lo...

Citation Context

...In data mining and machine learning, the study of real world networks is essential for the development of sound theoretical models, which ultimately lead to unbiased inferences and robust predictions =-=[35, 40, 44, 25, 26, 30, 36, 2, 45, 31, 3]-=-. Analyses of real networks rely upon two crucial tasks: (1) that of “generating,” or “sampling,” networks that display realistic properties of interest, and (2) that of “determining” which pure topol...

Citation Context

...ave become a central theme for KDD [29, 15, 18, 24, 28] On the other hand, in the social and mathematical sciences, (social and complex) networks have been an object of research for a few decades now =-=[19, 37, 10, 21, 6, 7, 12, 11]-=-. Over the years, the communication across communities has increased, the major results of each discipline have been shared and assimilated by the others, and, occasionally, old ideas have resurfaced ...

Citation Context

...In data mining and machine learning, the study of real world networks is essential for the development of sound theoretical models, which ultimately lead to unbiased inferences and robust predictions =-=[35, 40, 44, 25, 26, 30, 36, 2, 45, 31, 3]-=-. Analyses of real networks rely upon two crucial tasks: (1) that of “generating,” or “sampling,” networks that display realistic properties of interest, and (2) that of “determining” which pure topol...

Citation Context

...In data mining and machine learning, the study of real world networks is essential for the development of sound theoretical models, which ultimately lead to unbiased inferences and robust predictions =-=[35, 40, 44, 25, 26, 30, 36, 2, 45, 31, 3]-=-. Analyses of real networks rely upon two crucial tasks: (1) that of “generating,” or “sampling,” networks that display realistic properties of interest, and (2) that of “determining” which pure topol...

Citation Context

...nic texts have made available huge amount of “networked data,” to the data mining community at large, to the effect that models and algorithms for network analysis have become a central theme for KDD =-=[29, 15, 18, 24, 28]-=- On the other hand, in the social and mathematical sciences, (social and complex) networks have been an object of research for a few decades now [19, 37, 10, 21, 6, 7, 12, 11]. Over the years, the com...

Citation Context

...nic texts have made available huge amount of “networked data,” to the data mining community at large, to the effect that models and algorithms for network analysis have become a central theme for KDD =-=[29, 15, 18, 24, 28]-=- On the other hand, in the social and mathematical sciences, (social and complex) networks have been an object of research for a few decades now [19, 37, 10, 21, 6, 7, 12, 11]. Over the years, the com...

Citation Context

...eness of search strategies, e.g., greedy search finds short chains of acquaintances in small world networks [33, 34, 32, 1], at developing models of information flow [41, 42] and information exchange =-=[16]-=-, or at exploring the robustness of metrics for network analysis to variations in the topological properties [23, 8]. 3 Pure Topology Types We begin with a heuristic description of the pure topology t...

Citation Context

...nic texts have made available huge amount of “networked data,” to the data mining community at large, to the effect that models and algorithms for network analysis have become a central theme for KDD =-=[29, 15, 18, 24, 28]-=- On the other hand, in the social and mathematical sciences, (social and complex) networks have been an object of research for a few decades now [19, 37, 10, 21, 6, 7, 12, 11]. Over the years, the com...

Citation Context

...In data mining and machine learning, the study of real world networks is essential for the development of sound theoretical models, which ultimately lead to unbiased inferences and robust predictions =-=[35, 40, 44, 25, 26, 30, 36, 2, 45, 31, 3]-=-. Analyses of real networks rely upon two crucial tasks: (1) that of “generating,” or “sampling,” networks that display realistic properties of interest, and (2) that of “determining” which pure topol...

Citation Context

...nic texts have made available huge amount of “networked data,” to the data mining community at large, to the effect that models and algorithms for network analysis have become a central theme for KDD =-=[29, 15, 18, 24, 28]-=- On the other hand, in the social and mathematical sciences, (social and complex) networks have been an object of research for a few decades now [19, 37, 10, 21, 6, 7, 12, 11]. Over the years, the com...

Citation Context

...only used for network analysis. Typing network topologies form data is a fairly novel area of research. Initial explorations are present in specific application domains such as cover network analysis =-=[17]-=-. Related research efforts aim at providing intuitions and mathematical theory that describe what happens to topological properties when only partial information is available, e.g., sub-samples of sca...

Citation Context

...ee [43], at exploring the effectiveness of search strategies, e.g., greedy search finds short chains of acquaintances in small world networks [33, 34, 32, 1], at developing models of information flow =-=[41, 42]-=- and information exchange [16], or at exploring the robustness of metrics for network analysis to variations in the topological properties [23, 8]. 3 Pure Topology Types We begin with a heuristic desc...

Citation Context

... separability of the sampling algorithms surveyed or introduced in section 3.1. 2.1 Related Work The pure topology types we consider in the next section have been introduced separately over the years =-=[19, 38, 47, 5, 20, 9, 4, 39, 22]-=-. To the best of our knowledge neither exploratory nor compre1 The reference space used in this paper is defined by 47 metrics widely adopted in the social and physical sciences. We embed all sampled ...

Citation Context

...In data mining and machine learning, the study of real world networks is essential for the development of sound theoretical models, which ultimately lead to unbiased inferences and robust predictions =-=[35, 40, 44, 25, 26, 30, 36, 2, 45, 31, 3]-=-. Analyses of real networks rely upon two crucial tasks: (1) that of “generating,” or “sampling,” networks that display realistic properties of interest, and (2) that of “determining” which pure topol...

Citation Context

...34, 32, 1], at developing models of information flow [41, 42] and information exchange [16], or at exploring the robustness of metrics for network analysis to variations in the topological properties =-=[23, 8]-=-. 3 Pure Topology Types We begin with a heuristic description of the pure topology types. The sampling algorithms in the next section provide these intuitive notions with a precise meaning. Without lo...

Citation Context

...In data mining and machine learning, the study of real world networks is essential for the development of sound theoretical models, which ultimately lead to unbiased inferences and robust predictions =-=[35, 40, 44, 25, 26, 30, 36, 2, 45, 31, 3]-=-. Analyses of real networks rely upon two crucial tasks: (1) that of “generating,” or “sampling,” networks that display realistic properties of interest, and (2) that of “determining” which pure topol...

Citation Context

...ee [43], at exploring the effectiveness of search strategies, e.g., greedy search finds short chains of acquaintances in small world networks [33, 34, 32, 1], at developing models of information flow =-=[41, 42]-=- and information exchange [16], or at exploring the robustness of metrics for network analysis to variations in the topological properties [23, 8]. 3 Pure Topology Types We begin with a heuristic desc...

Citation Context

...local acquaintances, few of which relocate to places far away.” This stylized model of behavior is enough to replicate the phenomenon observed by Milgram, and it “sounds” like a plausible explanation =-=[38, 47]-=-. In section 3.1 we address the following problem. Problem 1 (Sampling) How can we generate topologies that have a set of desired properties with high probability? Sampling algorithms can be both dete...

Citation Context

...e introduce novel algorithms aimed at enhancing the diversity of sam5spled networks, and at addressing the case of cellular topology type. The following algorithms are available in C++ as part of ORA =-=[13]-=-. 3.1.1 Ring Lattice A ring lattice with parameters (n,k) is sampled as follows. Ring Lattice 1 Define n as the number of nodes in the graph, and k as the number of neighbors for each node. Given (n,k...