## A Framework of Filtering, Clustering and Dynamic Layout Graphs for Visualization (2005)

Venue: | ACSC |

Citations: | 1 - 0 self |

### BibTeX

@MISC{Huang05aframework,

author = {Xiaodi Huang and Peter Eades and Wei Lai},

title = {A Framework of Filtering, Clustering and Dynamic Layout Graphs for Visualization},

year = {2005}

}

### OpenURL

### Abstract

Many classical graph visualization algorithms have already been developed over the past decades. However, these algorithms face difficulties in practice, such as the overlapping node problem, large graph layout and dynamic graph layout. In order to solve these problems, this paper aims to systematically address algorithmic issues related to a novel framework that describes the process of graph visualization applications. First of all, a framework for graph visualization is described. As the important parts of this framework, we then develop two effective algorithms for filtering and clustering large graphs for the layouts. As for the dynamic graph layout, a new approach to removing overlapping nodes called force-transfer algorithm is developed. The framework has been implemented in a prototype called PGA to demonstrate the performance of the proposed algorithms. Finally, a case study is provided.

### Citations

798 | Matrix multiplication via arithmetic progressions - Coppersmith, Winograd - 1990 |

765 | The eyes have it: A task by data type taxonomy for information visualizations
- Shneiderman
- 1996
(Show Context)
Citation Context ...ovided. Keywords: information visualization, graph visualization, graph drawing, framework, filtering, clustering. 1 Introduction Many models have already been presented in information visualization (=-=Shneiderman 1996-=-, Haber et al 1990, Card et al 1999, Eades et al 2000, Chi 2000, Kreuseler et al 2002, Cristina et al 2003). These models focus on different aspects of visualization such as visualization process, des... |

471 |
Cluster Analysis
- Everitt, Landau, et al.
- 2001
(Show Context)
Citation Context ...are, the less link they are. For this reason, a linkage measure function is needed. The measures that occur most in the literature are the dot product, Euclidean distance and the Jaccard Coefficient (=-=Everitt 1993-=-). Among them, Jaccard Coefficient can be used to measure the degree of overlap. We therefore calculate the degree of the connection between two nodes l and k as 1 Link( l, k) = 1− deg( l) + deg( k) −... |

256 | A review and taxonomy of distortionoriented presentation techniques - Leung, Apperley - 1994 |

233 |
A new status index derived from sociometric analysis
- Katz
- 1953
(Show Context)
Citation Context ... have high rank scores, or by many others with low to moderate rank scores. The running time for computing NIS is O(n ^2.376) (Coppersmith 1990). This idea of eigenvector centrality was initiated by (=-=Katz 1953-=-) and further developed by (Hubbell 1965) and many others, finally culminated with (Bonacich 1972) who defined centrality as the principal eigenvector of an adjacency matrix. Up to this point, we alre... |

174 | Structural Analysis of Hypertexts: Identifying Hierarchies and Useful Metrics - Botafogo, Rivlin, et al. - 1992 |

129 | Visualization idioms: A conceptual model for scientific visualization systems - Haber, McNabb - 1990 |

117 | A method for comparing two hierarchical clusterings - Fowlkes, Mallows - 1983 |

92 | A Taxonomy of Visualization Techniques Using the Data State Reference Model. InfoVis '00 - Chi - 2000 |

80 | Multilevel visualization of clustered graphs - Eades, Feng - 1996 |

71 | Error-correcting isomorphisms of attributed relational graphs for pattern analysis - Tsai, Fu - 1979 |

52 | On maximum clique problems in very large graphs, in: External Memory Algorithms
- Abello, Pardalos, et al.
- 1999
(Show Context)
Citation Context ...phone billing records. The nodes of this “call graph” is telephone numbres, and the edges are calls made from one number from another. A one day call graph has 53,767,087 nodes and 170 million edges (=-=Abello et al 1999-=-). The Hollywood collaboration graph is the second example where the nodes are 225,000 actors, and an edges connects any two actors who have appeared in a feature film together. If we regard the World... |

46 | Navigating clustered graphs using force-directed methods - Eades, Huang |

37 |
An input-output approach to clique identification
- Hubbell
- 1965
(Show Context)
Citation Context ...hers with low to moderate rank scores. The running time for computing NIS is O(n ^2.376) (Coppersmith 1990). This idea of eigenvector centrality was initiated by (Katz 1953) and further developed by (=-=Hubbell 1965-=-) and many others, finally culminated with (Bonacich 1972) who defined centrality as the principal eigenvector of an adjacency matrix. Up to this point, we already know to how to calculate the NIS of ... |

34 | Algorithms for cluster busting in anchored graph drawing
- Lyons, Meijer, et al.
- 1998
(Show Context)
Citation Context ... nodeoverlapping problem have been reported in the literature: Uniform Scaling (Lai 1993), Constrained Optimization (He et al 1998, Marriott et all 2000) and Force-based Algorithms (Eades et al 1995, =-=Lyons et al 1998-=-, Lai 1993, 2002). Preserving the original structure of a graph, a straightforward approach called Uniform Scaling avoids node overlaps by uniformly scaling the overlapping layout. The nodes may, howe... |

30 | I.: An experimental study of the basis for graph drawing algorithms - PURCHASE, COHEN, et al. - 1997 |

28 |
Graph Drawing: Algorithms for the Visualisation of Graphs
- Battista, Eades, et al.
- 1999
(Show Context)
Citation Context ...aph applications in realistic application settings. From a practical perspective, we need to concentrate on questions such as what the steps for applying a large number of existing layout algorithms (=-=Battista et al 1999-=-) into real graphical applications are. In this paper, we propose a framework for practical graph visualization with the following purposes: • To establish a conceptual framework within which graph vi... |

28 | A flexible approach for visual data mining - Kreuseler, Schumann |

18 | Webofdav navigating and visualizing the web on-line with animated context swapping - Huang, Eades, et al. - 1998 |

15 |
Interactive Graph Layout: The Exploration of Large Graphs
- Henry
- 1992
(Show Context)
Citation Context ...ted edges. It is obvious that huang’s rules are concerned with both the structure and content of a Web graph. Simple filtering techniques based on node attributes are also reported in Henry’s thesis (=-=Henry 1992-=-). The above approaches have some limitations. The approach of Huang et al., for instance, is confined to a specific type of graph, namely Web graphs, and does not suffice to handle other types of gra... |

8 |
From Visual Data Exploration to Visual Data Mining: A Survey
- Oliveira, Levkowitz
- 2003
(Show Context)
Citation Context ...clustering. 1 Introduction Many models have already been presented in information visualization (Shneiderman 1996, Haber et al 1990, Card et al 1999, Eades et al 2000, Chi 2000, Kreuseler et al 2002, =-=Cristina et al 2003-=-). These models focus on different aspects of visualization such as visualization process, design, and guideline. In the following, we review such three typical models. Shneiderman (1996) pointed out ... |

8 | Tree visualization and navigation clues for information visualization - HERMAN, MELANCON - 1998 |

7 |
Structural and syntactic pattern recognition
- Bunke
- 1993
(Show Context)
Citation Context ...r to incorporate more semantic information about the properties of both objects and relations, attributed graphs (AGs) or attributed relational graph (ARG) were employed in many areas (Fu et al 1979, =-=Bunke 1993-=-, Bunke et al 1997, Lourens 1998, Walischewski 1997). Therefore, the problem of how to visualize attributed graphs should receive much more attention in the field of graph visualization (Huang 2001). ... |

4 |
Building Interactive Diagram Applications
- LAI
- 1993
(Show Context)
Citation Context ...le that adjustment of the original layout should be kept to a minimum. Roughly, three kinds of approaches to solving the nodeoverlapping problem have been reported in the literature: Uniform Scaling (=-=Lai 1993-=-), Constrained Optimization (He et al 1998, Marriott et all 2000) and Force-based Algorithms (Eades et al 1995, Lyons et al 1998, Lai 1993, 2002). Preserving the original structure of a graph, a strai... |

4 | EADES(2002): Removing edge-node intersections - LAI, PETER |

2 |
S.V.(2000): Graph Clustering by Flow Simulation Mathematics and
- Dongen
- 2000
(Show Context)
Citation Context ...fewer elements, allowing users to control the level of detail by “opening” and “closing” meta-nodes. The clustering approach has been taken by a number of graph drawing researchers (Eades et al 1996, =-=Dongen 2000-=-, Fowlkes et al 2001). Layout: Apply an algorithm to layout a given graph. View: With a layout, users may interact with the graph, changing the view in order to gain insight into the data. This featur... |

2 |
A biologically plausible model for corner based object recognition from colour images
- Lourens
- 1998
(Show Context)
Citation Context ... information about the properties of both objects and relations, attributed graphs (AGs) or attributed relational graph (ARG) were employed in many areas (Fu et al 1979, Bunke 1993, Bunke et al 1997, =-=Lourens 1998-=-, Walischewski 1997). Therefore, the problem of how to visualize attributed graphs should receive much more attention in the field of graph visualization (Huang 2001). A labeled graph, a special case ... |

2 | Indexing and retrieval of video using spatial reasoning techniques - Shearer - 1998 |

2 | MüLLER(1995): Graph layout adjustment strategies - STORY, HAUSI - 1995 |

1 |
Factoring and Weighting Approaches to Herman Status Scores and Clique Identification
- Bonacich
- 1972
(Show Context)
Citation Context ...for computing NIS is O(n ^2.376) (Coppersmith 1990). This idea of eigenvector centrality was initiated by (Katz 1953) and further developed by (Hubbell 1965) and many others, finally culminated with (=-=Bonacich 1972-=-) who defined centrality as the principal eigenvector of an adjacency matrix. Up to this point, we already know to how to calculate the NIS of a node in a graph. However, we cannot directly remove tho... |

1 | 2001): Information Visualization of Attributed Relational Data - Huang |

1 | Force-Transfer: A New Approach to Remove Node Overlapping in Graph Layout - Huang, Lai - 2003 |

1 |
Removing Node Overlapping
- Marriott, Stuckey, et al.
- 2000
(Show Context)
Citation Context ... initial and adjusted coordinates of the nodes within a graph. An optimal solution to such a function is then provided, subject to a set of constraints that ensure no node overlapping (He et al 1998, =-=Marriott et al 2000-=-). These kinds of approaches can “give better layout than the force scan algorithm, although they are slower”( Marriott et als2000). The Force-based Algorithm includes Cluster Busting in an anchored g... |