Shaojie Qiao

PutMode: prediction of uncertain trajectories in moving objects databases

Objective: Prediction of moving objects with uncertain motion patterns is emerging rapidly as a new exciting paradigm and is important for law enforcement applications such as criminal tracking analysis. However, existing algorithms for prediction in spatio-temporal databases focus on discovering frequent trajectory patterns from historical data. Moreover, these methods overlook the effect of some important factors, such as speed and moving direction. This lacks generality as moving objects may follow dynamic motion patterns in real life.Methods: We propose a framework for predicating uncertain trajectories in moving objects databases. Based on Continuous Time Bayesian Networks (CTBNs), we develop a trajectory prediction algorithm, called PutMode (Prediction of uncertain trajectories in Moving objects databases). It comprises three phases: (i) construction of TCTBNs (Trajectory CTBNs) which obey the Markov property and consist of states combined by three important variables including street identifier, speed, and direction; (ii) trajectory clustering for clearing up outlying trajectories; (iii) predicting the motion behaviors of moving objects in order to obtain the possible trajectories based on TCTBNs.Results: Experimental results show that PutMode can predict the possible motion curves of objects in an accurate and efficient manner in distinct trajectory data sets with an average accuracy higher than 80%. Furthermore, we illustrate the crucial role of trajectory clustering, which provides benefits on prediction time as well as prediction accuracy.

Discovering Organizational Structure in Dynamic Social Network

Applying the concept of organizational structure to social network analysis may well represent the power of members and the scope of their power in a social network. In this paper, we propose a data structure, called Community Tree, to represent the organizational structure in the social network. We combine the PageRank algorithm and random walks on graph to derive the community tree from the social network. In the real world, a social network is constantly changing. Hence, the organizational structure in the social network is also constantly changing. In order to present the organizational structure in a dynamic social network, we propose a tree learning algorithm to derive an evolving community tree. The evolving community tree enables a smooth transition between the two community trees and well represents the evolution of organizational structure in the dynamic social network. Experiments conducted on real data show our methods are effective at discovering the organizational structure and representing the evolution of organizational structure in a dynamic social network.

SimRank: A Page Rank approach based on similarity measure

As the Web contains rich and convenient information, Web search engine is increasingly becoming the dominant information retrieving approach. In order to rank the query results of web pages in an effective and efficient fashion, we propose a new page rank algorithm based on similarity measure from the vector space model, called SimRank, to score web pages. Firstly, we propose a new similarity measure to compute the similarity of pages and apply it to partition a web database into several web social networks (WSNs). Secondly, we improve the traditional PageRank algorithm by taking into account the relevance of page to a given query. Thirdly, we design an efficient web crawler to download the web data. And finally, we perform experimental studies to evaluate the time efficiency and scoring accuracy of SimRank with other approaches.

Optimal Multiple Sink Nodes Deployment in Wireless Sensor Networks Based on Gene Expression Programming

In wireless sensor networks (WSN) data transmission is usually performed by sensors in manner of multi-hop forwarding towards a central static control center (sink). A lot of cheap, low-powered and energy-limited sensors are deployed in the monitored area and some of these nodes closer to the sink node use up their energy more quickly than other nodes because they relay more packets. Although most of the sensor nodes have enough energy left to work, the energy consumption imbalance leads to connectivity holes and coverage holes, and finally the whole network failure. The main contributions of this paper include: (a) a new scheme based on multiple sink nodes is proposed to prolong the network lifetime and to reduce the response time. It is effective, especially in the target tracking applications, (b) the deployment strategy with given number of multiple sink nodes is explored in the grid sensor network, (c) Gene Expression Programming based Multiple Sink Nodes deployment algorithm (GEP-MSN) is proposed to optimally deploy multiple sink nodes over the monitored region, (d) a data transmission cost model (TCM) is introduced to measure the cost for optimizing during the transmission phase, (e) extensive simulations are conducted to show that the scheme can greatly extend the network lifetime by around 16.6% and 36.3% on average compared with two naive methods based on random distributed sink nodes.

A new blockmodeling based hierarchical clustering algorithm for web social networks

Cluster analysis for web social networks becomes an important and challenging problem because of the rapid development of the Internet community like YouTube, Facebook and TravelBlog. To accurately partition web social networks, we propose a hierarchical clustering algorithm called HCUBE based on blockmodeling which is particularly suitable for clustering networks with complex link relations. HCUBE uses structural equivalence to compute the similarity among web pages and reduces a large and incoherent network into a set of smaller comprehensible subnetworks. HCUBE is actually a bottom-up agglomerative hierarchical clustering algorithm which uses the inter-connectivity and the closeness of clusters to group structurally equivalent pages in an effective fashion. In addition, we address the preliminaries of the proposed blockmodeling and the theoretical foundations of HCUBE clustering algorithm. In order to improve the efficiency of HCUBE, we optimize it by reducing its time complexity from O(|V|^2) to O(|V|^2/p), where p is a constant representing the number of initial partitions. Finally, we conduct experiments on real data and the results show that HCUBE is effective at partitioning web social networks compared to the Chameleon and k-means algorithms.

Clustering Without Prior Knowledge Based on Gene Expression Programming

Most existing clustering methods require prior knowledge, such as the number of clusters and thresholds. They are difficult to determine accurately in practice. To solve the problem, this study proposes a novel clustering algorithm named GEP-Cluster based on Gene Expression Programming (GEP) without prior knowledge. The main contributions include: (1) a new concept named Clustering Algebra is proposed that makes clustering as algebraic operation , (2) a GEP-Cluster algorithm is proposed to find the best clustering information automatic by GEP and discover the best clustering solution without any prior knowledge, (3) an AMCA (Automatic Merging Cluster Algorithm) algorithm is proposed to merge clustering automatically. Extensive experiments demonstrate that GEP-Cluster algorithm is effective in clustering without any prior knowledge on various data sets.

PartSpan: Parallel Sequence Mining of Trajectory Patterns

The trajectory pattern mining problem has recently attracted increasing attention. This paper precisely addresses the parallel mining problem of trajectory patterns as well as the newly proposed concepts with regard to trajectory pattern mining. An efficient parallel trajectory sequential pattern mining (PartSpan) is proposed by incorporating three key techniques: prefix-projection, parallel formulation, and candidate pruning. The prefix-projection technique is used to decompose the search space as well as greatly reducing candidate trajectory sequences. The parallel formulation integrates the data parallel formulation and the task parallel formulation to partition the computations and to assign them to multiple processors in an efficient and effective manner that helps reduce the communication cost across processors. Representative experiments are used to evaluate the performance of PartSpan. The results show that PartSpan outperforms GSP-based and SPADE-based parallel algorithms in mining very large trajectory databases.

Constrained k-closest pairs query processing based on growing window in crime databases

Spatial analysis in crime databases has recently been an active research topic. To solve the problem of finding the closest pairs of objects within a given spatial region, as required in crime geo-data applications, this paper proposes an efficient constrained k-closest pairs query processing algorithm based on growing window. It expands the window gradually instead of searching the whole workspace for multiple types of spatial objects. It employs a density-based range estimation approach to calculate the square query range and an optimized R-tree to store the index entities. In addition, a distance threshold T for the closest pair of objects is introduced to prune tree nodes. Experiments evaluate the effect of three important factors, i.e., the portion of overlapping between the workspaces of two data sets, the value of k, and the size of buffer. The results show that the new algorithm outperforms the heap-based approach.

VCCM mining: mining virtual community core members based on gene expression programming

Intelligence operation against the terrorist network has been studied extensively with the aim to mine the clues and traces of terrorists. The contributions of this paper include: (1) introducing a new approach to classify terrorists based on Gene Expression Programming (GEP); (2) analyzing the characteristics of the terrorist organization, and proposing an algorithm called Create Virtual Community (CVC) based on tree-structure to create a virtual community; (3) proposing a formal definition of Virtual Community (VC) and the VCCM Mining algorithm to mine the core members of a virtual community. Experimental results demonstrate the effectiveness of VCCM Mining.

WebRank: a hybrid page scoring approach based on social network analysis

Applying the centrality measures from social network analysis to score web pages may well represent the essential role of pages and distribute their authorities in a web social network with complex link structures. To effectively score the pages, we propose a hybrid page scoring algorithm, called WebRank, based on the PageRank algorithm and three centrality measures including degree, betweenness, and closeness. The basis idea of WebRank is that: (1) use PageRank to accurately rank pages, and (2) apply centrality measures to compute the importance of pages in web social networks. In order to evaluate the performance of WebRank, we develop a web social network analysis system which can partition web pages into distinct groups and score them in an effective fashion. Experiments conducted on real data show that WebRank is effective at scoring web pages with less time deficiency than centrality measures based social network analysis algorithm and PageRank.