#### DMCA

## Cascading spatio-temporal pattern discovery

Citations: | 12 - 5 self |

### Citations

14035 |
Computers and intractability: a guide to the theory of NP-completeness, Freeman,
- Garey, Johnson
- 1979
(Show Context)
Citation Context ... identify frequent sub-graph patterns from a large sparse graph using computationally expensive measures such as the Maximum Independent set (MIS) [15]. The problem of computing an MIS is NP-complete =-=[12, 15]-=-. In addition, a statistical/probabilistic interpretation of MIS has not been explored. A special case of large sparse graph mining is workflow process mining that deals with finding a minimal directe... |

8894 |
Probabilistic Reasoning in Intelligent Systems
- Pearl
- 1988
(Show Context)
Citation Context ...hods are not designed to discover ST partial ordered patterns such as CSTPs. In the broader data mining literature, possible candidates for quantifying the interestingness of CSTPs have been proposed =-=[15, 14, 19]-=-. Section 5 discusses their limitations in a ST data mining context. Our Contributions: This paper models CSTPs as partial ordered ST patterns. A novel CSTP interest measure, the Cascade Participation... |

3326 | Mining association rules between sets of items in large databases
- Agrawal, Imielinski, et al.
- 1993
(Show Context)
Citation Context ...ex (CPI) which can be evaluated in O(n2 )computations(nbeing the number of instances in the database) is proposed. The CPI also exhibits the anti-monotone property to facilitate apriori style pruning =-=[3]-=-. The CPI is an upper bound to the space-time K function [21, 9]. The paper introduces a novel CSTP miner and proves that it is correct and complete. In addition to apriori style pruning and upper bou... |

2936 | Maintaining Knowledge about Temporal Intervals
- Allen
(Show Context)
Citation Context ...of parameters using synthetic datasets and evaluate alternatives to the CPI. We hope to investigate new interest measures that for account aspects such as scale and ST semantics (e.g., time intervals =-=[11, 4]-=-). Finally, Based on ST patterns from applications such as spatial epidemiology [7], spatial economics [10] and chemical morphogenesis [27], we plan to explore guidelines to identify neighborhood size... |

1037 |
Towards a general theory of action and time
- Allen
(Show Context)
Citation Context ...her disjoint or similar occurrence times. This limits the topological richness of ST patterns to account for notions such as events and processes that are defined in time geography and temporal logic =-=[5, 24, 29]-=-. Hence, existing ST data mining methods are not designed to discover ST partial ordered patterns such as CSTPs. In the broader data mining literature, possible candidates for quantifying the interest... |

757 | Mining sequential patterns: Generalizations and performance improvements
- Srikant, Agrawal
- 1996
(Show Context)
Citation Context ...conditional probability of a CSTP given one of its participating event types. Other alternatives to quantify interestingness have been explored in the broader data mining literature [19], [20], [21], =-=[22]-=-. For example, transaction based frequent pattern discovery methods for extracting sequences and graphs seek to identify a set of frequent patterns given a set of transactions from marketbasket data o... |

335 |
The spatial economy. Cities, regions and international trade
- Fujita, Krugman, et al.
- 1999
(Show Context)
Citation Context ...st measures that for account aspects such as scale and ST semantics (e.g., time intervals [11, 4]). Finally, Based on ST patterns from applications such as spatial epidemiology [7], spatial economics =-=[10]-=- and chemical morphogenesis [27], we plan to explore guidelines to identify neighborhood sizes and compare patterns with those generated using Graphical models like Bayesian networks. 7 Acknowledgment... |

335 |
The kolmogorov-smirnov test for goodness of fit
- Massey
- 1951
(Show Context)
Citation Context ...WLEDGE AND DATA ENGINEERING, VOL. , NO., 29 tests. To test the significance of this pattern, we performed the Kolmogorov-Smirnov (KS) nonparameteric test for equality of two statistical distributions =-=[18]-=-. The null hypothesis under the KS test states that the two statistical samples being compared are from the same underlying population. The rejection of the null hypothesis implies that the two sample... |

264 | Mining Process Models from Workflow Logs”,
- Agrawal, Gunopulos, et al.
- 1998
(Show Context)
Citation Context ...se of large sparse graph mining is workflow process mining that deals with finding a minimal directed acyclic graph of a given process and a log containing many independent executions of this process =-=[2]-=-. This approach is not suitable for CSTP mining due to potential overlap among CSTP instances and the presence of multiple types of CSTPs in a dataset. Models such as Bayesian networks have been used ... |

231 | Query Optimization in Database Systems, in:
- Jarke, Koch
- 1984
(Show Context)
Citation Context ...est measures for exploring these type of patterns since CPI may not retain its antimonotonicty. In addition, we will explore alternatives to the current algorithms using ideas from query optimization =-=[31]-=-. This is because, a size-3 pattern (e.g. B → A, B → C, A → C ) may be computed by joining instances of either (B → A, B → C) and (B → C, A → C) or (B → A, B → C) and (B → A, A → C) or (B → C, A → C) ... |

184 | Partition Based Spatial-Merge Join,
- Patel, DeWitt
- 1996
(Show Context)
Citation Context ...join operator between CSTPs of different sizes. Computing an ST join can be performed using several techniques. We describe a new algorithm for achieving this: a spatio-temporal partition based CSTPM =-=[16]-=-. Figure 7 shows a block diagram overview of the two different CSTPM April 29, 2011 DRAFTIEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, VOL. , NO., 14 algorithms. In our recent work presented a... |

177 |
Modeling spatial patterns.
- Ripley
- 1977
(Show Context)
Citation Context ... can be proved to be antimonotonic and an upper bound to a ST statistical measure called the space-time K-Function, which is a generalization of the spatial statistical measure Ripley’s K [13], [14], =-=[15]-=-. Solutions to different challenges are based on two principles: (1) prevent unneccessary interest measure computation and (2) speed up interest measure computation. A natural way to deal with the exp... |

127 | Finding frequent patterns in a large sparse graph
- Kuramochi, Karypis
- 2004
(Show Context)
Citation Context ...hods are not designed to discover ST partial ordered patterns such as CSTPs. In the broader data mining literature, possible candidates for quantifying the interestingness of CSTPs have been proposed =-=[15, 14, 19]-=-. Section 5 discusses their limitations in a ST data mining context. Our Contributions: This paper models CSTPs as partial ordered ST patterns. A novel CSTP interest measure, the Cascade Participation... |

120 | An efficient algorithm for discovering frequent subgraphs
- Kuramochi, Karypis
- 2004
(Show Context)
Citation Context ...hods are not designed to discover ST partial ordered patterns such as CSTPs. In the broader data mining literature, possible candidates for quantifying the interestingness of CSTPs have been proposed =-=[15, 14, 19]-=-. Section 5 discusses their limitations in a ST data mining context. Our Contributions: This paper models CSTPs as partial ordered ST patterns. A novel CSTP interest measure, the Cascade Participation... |

73 | Discovering spatial co-location patterns: a summary of results
- Shekhar, Huang
(Show Context)
Citation Context ...pating event type. In order to account for the unique characteristics of ST frameworks, the measures for CSTP mining are a generalization of the measures defined in spatial co-location pattern mining =-=[23]-=-. Definition 2.1. A Cascade Participation Ratio, CPR(CSTP,M), may be interpreted as an estimate of the conditional probability of an instance of a pattern CSTP given an instance of event type M, i.e. ... |

69 |
The challenge of emerging and re-emerging infectious diseases.
- Morens, Folkers, et al.
- 2004
(Show Context)
Citation Context ...ncluding climate change science (e.g. understanding the effects of climate change on food supply[1]), public health (e.g. tracking the emergence, spread and reemergence of multiple infectious diseases=-=[18]-=-) and public safety (e.g. studying the ST patterns of different crime generators). CSTP discovery is a challenging problem for two key reasons: (1) quantifying the measure of interestingness of ST pat... |

61 | Discovering significant patterns
- Webb
(Show Context)
Citation Context ... verifying actionable patterns. Hence, a key component of future work is to perform computationally expensive significance tests by randomly permuting ST instances and accounting for multiple testing =-=[30]-=-. Since the MST filter is sensitive to grid cell sizes, we will explore ways to determine practical grid cell sizes in collaboration with domain scientists. In addition, we will explore new interest m... |

57 | Discovering colocation patterns from spatial datasets: A general approach
- Huang, Shekhar, et al.
- 2004
(Show Context)
Citation Context .... 2. Illustrative ST crime dataset and cascading ST pattern patterns (e.g. spatial co-locations and topological patterns) represent unordered subsets of spatial or ST events in a uniform ST framework =-=[5]-=-, [6]. Totally ordered patterns (e.g. ST sequences) represent a linear chain reaction of ST event types [7]. In a partially ordered set, for some (but possibly not all) pairs of elements, one of the e... |

57 | Event-oriented approaches to geographic phenomena
- Worboys
- 2005
(Show Context)
Citation Context ... of which generally represent change of some kind. Processes refer to ongoing phenomena that represent activities of one or more types without a specific endpoint, such as global climate change [11], =-=[12]-=-. Events refer to individual occurrences of a process with a specific beginning and end. Event-types and event-instances are distinguished. For example, a hurricane event may occur at many different l... |

29 | Towards a qualitative theory of movement
- Galton
- 1995
(Show Context)
Citation Context ...of parameters using synthetic datasets and evaluate alternatives to the CPI. We hope to investigate new interest measures that for account aspects such as scale and ST semantics (e.g., time intervals =-=[11, 4]-=-). Finally, Based on ST patterns from applications such as spatial epidemiology [7], spatial economics [10] and chemical morphogenesis [27], we plan to explore guidelines to identify neighborhood size... |

26 | Discovery of periodic patterns in spatio- temporal sequences.
- Cao, Mamoulis, et al.
- 2007
(Show Context)
Citation Context ...TP instances and the presence of multiple types of CSTPs in a dataset. Time snapshot based spatial pattern families are primarily drawn from two areas: (1) moving object database patterns [26], [27], =-=[28]-=- and (2) dynamic graph analysis [29]. Time snapshot based spatial pattern families deal with space and time separately and perform a partitioning over time. Such a partitioning may be meaningful in a ... |

24 |
Second-order analysis of space-time clustering.
- Diggle, Chetwynd, et al.
- 1995
(Show Context)
Citation Context ... the number of instances in the database) is proposed. The CPI also exhibits the anti-monotone property to facilitate apriori style pruning [3]. The CPI is an upper bound to the space-time K function =-=[21, 9]-=-. The paper introduces a novel CSTP miner and proves that it is correct and complete. In addition to apriori style pruning and upper bound (UB) filtering, the CSTP miner uses the ST nature of the data... |

24 |
Counting labeled acyclic digraphs, New Directions in Graph Theory,
- Robinson
- 1973
(Show Context)
Citation Context ...statistical interpretation is based on ST statistics) and (2) the large cardinality of candidate patterns, which is exponential in the number of event types, makes the problem combinatorially complex =-=[22]-=-. Related Work : Related literature from ST data mining has primarily focussed on ST sequences [13] and un-ordered co-occurrences [28, 6]. A ST sequence represents a chain of event types in a uniform ... |

19 |
Spatial diffusion: An Historical Geography of Epidemics in an Island Community.
- Cliff, Haggett, et al.
- 1981
(Show Context)
Citation Context ... investigate new interest measures that for account aspects such as scale and ST semantics (e.g., time intervals [11, 4]). Finally, Based on ST patterns from applications such as spatial epidemiology =-=[7]-=-, spatial economics [10] and chemical morphogenesis [27], we plan to explore guidelines to identify neighborhood sizes and compare patterns with those generated using Graphical models like Bayesian ne... |

17 | A framework for mining sequential patterns from spatio-temporal event data sets
- Huang, Zhang, et al.
- 2008
(Show Context)
Citation Context ...tterns, which is exponential in the number of event types, makes the problem combinatorially complex [22]. Related Work : Related literature from ST data mining has primarily focussed on ST sequences =-=[13]-=- and un-ordered co-occurrences [28, 6]. A ST sequence represents a chain of event types in a uniform ST framework under the assumptions of total ordering because of time [13]. Co-occurrences represent... |

15 |
Discovery of collocation episodes in spatiotemporal data
- Cao, Mamoulis, et al.
- 2006
(Show Context)
Citation Context ...lap among CSTP instances and the presence of multiple types of CSTPs in a dataset. Time snapshot based spatial pattern families are primarily drawn from two areas: (1) moving object database patterns =-=[26]-=-, [27], [28] and (2) dynamic graph analysis [29]. Time snapshot based spatial pattern families deal with space and time separately and perform a partitioning over time. Such a partitioning may be mean... |

14 | Mixed-drove spatiotemporal co-occurrence pattern mining.
- Celik, Shekhar, et al.
- 2008
(Show Context)
Citation Context ... number of event types, makes the problem combinatorially complex [22]. Related Work : Related literature from ST data mining has primarily focussed on ST sequences [13] and un-ordered co-occurrences =-=[28, 6]-=-. A ST sequence represents a chain of event types in a uniform ST framework under the assumptions of total ordering because of time [13]. Co-occurrences represent un-ordered collection of event types ... |

14 |
A framework for mining topological patterns in spatio-temporal databases
- Wang, Hsu, et al.
- 2005
(Show Context)
Citation Context ...Illustrative ST crime dataset and cascading ST pattern patterns (e.g. spatial co-locations and topological patterns) represent unordered subsets of spatial or ST events in a uniform ST framework [5], =-=[6]-=-. Totally ordered patterns (e.g. ST sequences) represent a linear chain reaction of ST event types [7]. In a partially ordered set, for some (but possibly not all) pairs of elements, one of the elemen... |

13 | Discovering correlated spatio-temporal changes in evolving graphs
- Chan, Bailey, et al.
- 2008
(Show Context)
Citation Context ...tiple types of CSTPs in a dataset. Time snapshot based spatial pattern families are primarily drawn from two areas: (1) moving object database patterns [26], [27], [28] and (2) dynamic graph analysis =-=[29]-=-. Time snapshot based spatial pattern families deal with space and time separately and perform a partitioning over time. Such a partitioning may be meaningful in a moving object database setting (e.g.... |

5 |
et al., Discovering frequent closed partial orders from strings
- Pei, Wang, et al.
- 2006
(Show Context)
Citation Context ...ds for extracting sequences and graphs seek to identify a set of frequent patterns given a set of transactions from market-basket data or other graph structure transactions such as chemical compounds =-=[14, 25, 20]-=-. These methods use support (probability of occurrence) to denote the interestingness of a pattern. However, ST frameworks are continuous. Transactionization/partitioning of a continuous framework mis... |

4 |
city crime records. http://www.lincoln.ne.gov/city/police
- Lincoln
- 2008
(Show Context)
Citation Context ... Figure 6: The CPI as an upper bound to the space-time K-Function[21] 4 Case Study and Performance Evaluation In this section, we present a case study using real crime datasets from Lincoln, Nebraska =-=[8]-=- and a computational performance evaluation of the CSTP miner. 1 1 C.4 A.4 Space CSTP Instances from coarse dataset B A B C (1,1) (0,1) (1,1) (2,1) (3,0) (2,1) (3,0) (3,1) 1 (1,1) (0,2) (1,1) (2,2) (3... |

2 | Assaults in and around bars(2nd edition
- Scott, Dedel
- 2006
(Show Context)
Citation Context ...al instances of the CSTP in Figure 2(a) are shown as three striped triangles in Figure 3. Bars in large cities are often considered as potential generators of crime that occurs after bar closing time =-=[1]-=-. In crime analysis, CSTPs may suggest interesting hypotheses relating several crime types, which may help law enforcement agencies, public safety groups and policy makers to determine appropriate act... |

2 | Discovering partially ordered patterns of Terrorism via - Shine, Rogers, et al. - 2010 |