#### DMCA

## Profiling Social Network Users with Machine Learning

### Citations

11951 | Maximum likelihood from incomplete data via the em algorithm
- Dempster, Laird, et al.
- 1977
(Show Context)
Citation Context ...ers, use them to calculate the cluster probabilities for each instance, use these probabilities to reestimate the parameters, and repeat. This is called the EM algorithm, for expectation–maximization =-=[18]-=-. The first step, calculation of the cluster probabilities (which 4. DECISION TREES One of the most successful models in machine learning are decision trees. Decision tree learning is a method for app... |

6599 | C4.5: programs for machine learning - QUINLAN - 1993 |

4371 | Induction of decision trees
- Quinlan
- 1986
(Show Context)
Citation Context ...s are variations on a core algorithm that employs a top-down, greedy search through the space of possible decision trees. This approach is exemplified by the ID3 algorithm [18] and its successor C4.5 =-=[19]-=-. ID3 learns decision trees by constructing them top-down, beginning at the root of the tree and deciding which attribute should be tested. To perform this decision, each instance attribute is evaluat... |

206 | Microscopic evolution of social networks
- Leskovec, Backstrom, et al.
- 2008
(Show Context)
Citation Context ...challenging task in machine learning. This task is mainly concerned with modeling and often discovering the dynamics of the social graph. An interesting work in this direction is the one presented in =-=[8]-=- where the authors present a detailed study of network evolution by analyzing four large online social networks. They exploit the full temporal information about node and edge arrivals. This study per... |

164 | Feedback effects between similarity and sociai infiuence in online communities. In:
- Crandall, Cosley, et al.
- 2008
(Show Context)
Citation Context ...scovery of topic diffusion and evolution benefits from this joint inference, and the probabilistic model proposed performs significantly better than existing methods. Another approach is presented in =-=[15]-=-, where the authors develop techniques for identifying and modeling the interactions between social influence and selection, using data from online communities where both social interaction and change... |

62 |
Finding, Counting and Listing all Triangles in Large Graphs, An Experimental Study.
- Schank, Wagner
- 2005
(Show Context)
Citation Context ...s, computing the global clustering coefficient can become computationally intractable if we rely on exact counting. The exact counting of triangles has been shown to be computationally very expensive =-=[4, 5]-=-. Other approaches base the counting on approximations such as the work presented in [6]. Recently in [7] the authors presented a label propagation approach to community structure discovery. They intr... |

44 | Main-memory triangle computations for very large (sparse (power-law)) graphs.
- Latapy
- 2008
(Show Context)
Citation Context ...s, computing the global clustering coefficient can become computationally intractable if we rely on exact counting. The exact counting of triangles has been shown to be computationally very expensive =-=[4, 5]-=-. Other approaches base the counting on approximations such as the work presented in [6]. Recently in [7] the authors presented a label propagation approach to community structure discovery. They intr... |

22 | Transitive node similarity for link prediction in social networks with positive and negative links.
- Symeonidis, Tiakas, et al.
- 2010
(Show Context)
Citation Context ...d to each other but have in common other users. 5.2 Preprocessing and input engineering The data used in our experiment have been extracted from the online social network Hi5 1 and first presented in =-=[21]-=-. This dataset is composed of 4928 users where for every user there are 20 other users considered as his closest links. In Figure 2 it is shown the table of users where every user is in a row and the ... |

21 | Mining periodic behavior in dynamic social networks
- Lahiri, Berger-Wolf
- 2008
(Show Context)
Citation Context ...ases. Social interactions that occur regularly will typically correspond to significant yet often infrequent and hard to detect interaction patterns. To identify such regular behavior, the authors in =-=[9]-=- propose a new mining problem of finding periodic or near periodic subgraphs in dynamic social networks where scalability is also a major issue. They propose a practical, efficient and scalable algori... |

19 | Graph Mining Applications to Social Network Analysis.
- Tang, Liu
- 2010
(Show Context)
Citation Context ...a friend are very likely to be also friends. An interesting coefficient is the proposed approach to measure the transitivity as the probability of connections between one vertex's neighboring friends =-=[3]-=-. For real social networks, computing the global clustering coefficient can become computationally intractable if we rely on exact counting. The exact counting of triangles has been shown to be comput... |

16 |
A scalable framework for modeling competitive diffusion in social networks.
- Broecheler, Shakarian, et al.
- 2010
(Show Context)
Citation Context ...analyzed in short time scales. Modeling event propagation is another important challenge in social networks. Handling this task appropriately leads to interesting applications for viral marketing. In =-=[13]-=-, the authors propose a scalable framework for modeling competitive diffusion in social networks. In social networks, multiple phenomena often diffuse in competition with one another. Applications of ... |

14 |
The joint inference of topic diffusion and evolution in social communities.
- Lin, Mei, et al.
- 2011
(Show Context)
Citation Context ...very of the diffusion paths and the evolutionary process of a topic. Unlike explicit user behavior (e.g., buying a book) both these are implicit. An interesting approach has been recently proposed in =-=[14]-=- where the authors track the evolution of an arbitrary topic and reveal the latent diffusion paths of that topic in a social community. The proposed approach is based on a novel and principled probabi... |

10 | Top-K aggregation queries over large networks.
- Yan, He, et al.
- 2010
(Show Context)
Citation Context ...implementation of aggregation operations on relational databases does not guarantee superior performance in network space, especially when it involves edge traversals and joins of gigantic tables. In =-=[22]-=-, the authors investigate the neighborhood aggregation queries: Find nodes that have top-k highest aggregate values over their h-hop neighbors. While these basic queries are common in a wide range of ... |

3 |
L.: Label propagation algorithm: a semi-synchronous approach
- Cordasco, Gargano
- 2012
(Show Context)
Citation Context ...unting. The exact counting of triangles has been shown to be computationally very expensive [4, 5]. Other approaches base the counting on approximations such as the work presented in [6]. Recently in =-=[7]-=- the authors presented a label propagation approach to community structure discovery. They introduce a semi–synchronous version of label propagation algorithms which aims to combine the advantages of ... |

3 |
Linkboost: A novel cost-sensitive boosting framework for community-level network link prediction
- Comar, Tan, et al.
- 2011
(Show Context)
Citation Context ...from the whole network. An interesting approach is community (cluster) level link prediction method without the need to explicitly identify the communities in a network. This approach is presented in =-=[11]-=- where the authors define a variable-cost loss function to address the data skewness problem. They provide theoretical proof that shows the equivalence between maximizing the well-known modularity mea... |

1 |
A Review of Machine Learning and Data Mining Approaches for Business Applications in Social Networks
- Trandafili, Biba
- 2013
(Show Context)
Citation Context ...ng 1. INTRODUCTION Social Network Mining (SNM) is a rapidly growing field that is increasingly receiving much attention in both the communities of data mining and in marketing and business strategies =-=[1]-=-. The main goal of this scientific area is the study of relationships between individuals regarding their social position, the analysis of their roles, the discovery of social structures and many othe... |

1 | Scalable and High Performing Learning and Mining in Large-Scale Networked Environments: A State-of-the-art Survey - Trandafili, Biba - 2013 |

1 |
2010. Analysis of Large Multi-modal Social Networks: Patterns and a Generator
- Ch
(Show Context)
Citation Context ...le usually construct an explicit social network by adding each other as friends, but they can also build implicit social networks through daily actions like commenting on posts, or tagging photos. In =-=[10]-=- it is addressed this problem: given a real social networking system which changes over time, do daily interactions follow any pattern? The authors model the formation and co-evolution of multi-modal ... |

1 |
Prediction Based on Subgraph Evolution in Dynamic Social Networks
- Juszczyszyn, Musial, et al.
(Show Context)
Citation Context ...ions and aggregating the weak learners constructed from each partition. The authors empirically evaluate the proposed algorithm by evaluating it on 4 real-world network datasets. In a recent approach =-=[12]-=-, the authors proposed a new method for characterizing the dynamics of complex networks with an application to the link prediction problem. The approach proposed is based on the discovery of network s... |