Junzhong Gu

Privacy Preserving Reverse Nearest-Neighbor Queries Processing on Road Network

In recent years, with the popularity of Location Based Service (LBS) and recommendation system, spatial data query has become a hot study area. Reverse nearest neighbor (RNN) query is one of the most important queries in spatial database. It plays an important role in decision-making system, recommended system and frameworks like so on. In many cases, users do not want to disclose the specific location information to the system. It requires a certain extent anonymous of user information. Compared to the common Euclidean space, Road Network is more practical. However, in previous studies, there is no RNN queries base on road network taking into account the protection of user privacy. In this paper, we propose a novel algorithm- RN-BRNN(Road Network - Bichromatic Reverse Nearest Neighbor) query algorithm, which considering both the road network bichromatic RNN query and user location privacy protection. RN-BRNN algorithm establishes a special Voronoi Cell based on the road network, queried points anonymity, and probability calculus of obtained RNN. Extensive experimental results show that the algorithm maintenances the same time-complexity with the Euclidean space, and improved precision greatly.

SOCOM: Multi-sensor Oriented Context Model Based on Ontologies

Sensor plays an important role in context-aware computing. While sensor modeling is usually isolated from former researches on context modeling and sensor type is always restricted to physical ones, this paper aims to provide a more comprehensive insight into the relationships between sensor and context. Based on a more generic definition of sensor and a corresponding sensor category in context-aware computing, main characteristics of sensor and the relations between sensor and context are analyzed in details. In particular, a multi-sensor oriented context model based on ontology approach is proposed. The model which has been applied to a middleware supporting for rapid context-aware applications development is proved to be effective.

Link Prediction Using Protected Location History

This paper investigates the link prediction problem in location-based social networking services (LBSNS) with protected location history. While former approaches mainly utilize the accurate locations, the relevant data we analyzed are modeled by a location privacy protection model called k-anonymous spatial-temporal cloaking model (KSTCM) which perturbs the location-related records on both temporal and spatial dimensions. We also propose a KSTCM-based co-located relationship model based on which various contextual features are extracted to help building the prediction model. Furthermore, we study the extent to which link between two users can be inferred from the co-located situation in space, time and degree of privacy protection. The experimental results show that our link prediction model can obtain a very high accuracy.

Metadata management of context resources in context-aware middleware system

Metadata is the data about data. With the development of information technologies, it is realized that metadata is of great importance in discovery, identification, localization and access of resources. For the context-aware middleware, it is necessary to provide the metadata of context resources connected to it for the context consumers. However, the complexity of context-aware computing results in the diversity of context resources which makes it difficult in explicitly describing the metadata. In this paper, an XML-based markup language called VirtualSensorML is proposed for the purpose of providing model and schema to describe the metadata of context resources in an understandable format. It defines fundamental elements and various sub elements to describe the common and special properties of different context resources. The metadata management of context resources is currently implemented in a multi-agent based context-aware middleware. The evaluation demonstrates that it is efficient to improve the discovery of context resources.

Continuous Skyline Queries with Integrity Assurance in Outsourced Spatial Databases

Integrity assurance is an important problem for query processing in outsourced spatial databases, where the location-based service (LBS) provides query services to the clients on behalf of the data owner. If the LBS server is not trustworthy, it may return incorrect or incomplete query results intentionally or unintentionally. Therefore, to ensure the query integrity, the data owner needs to build additional authenticated data structures so that the clients can authenticate the soundness and completeness of query results. In this paper, we study the integrity assurance problem for continuous location-based skyline queries. We propose three novel techniques based on MR-Sky-tree, i.e., using valid scope, visible region, and incremental VO to reduce the computation and communication cost. Experimental results show that our proposed techniques achieve shorter computation time and lower communication cost than the existing approach.

Dependency Tree Based Chinese Relation Extraction over Web Data

A new semi-supervised approach for Chinese relation extraction (RE) over constantly growing and edgeless web data is introduced in this paper. Existing semi-supervised approaches have the better improvement potential while lacking syntactic structure and semantic meaning of a sentence and unsuitable to loosely structured Chinese sentences. To follow their basic procedures as well as covering their remaining shortages, a dependency tree (DT) including both structure and semantic information is drawn in. Based on DTs, a new kind of pattern, called DT-based pattern, is proposed to extract new triples. Later patterns are optimized according to the characteristics of Chinese and typed dependency trees. Finally, extensive experiments show the higher precision and more efficiency of the proposed approach against DIPRE.

Region of Interest Discovery in Location-Based Social Networking Services with Protected Locations

Region of Interest (ROI) discovery is one of the most common interests in Location-based social networking services (LBSNS). While former researches mainly utilize the accurate location history, this paper explores the methods to extract those regions with protected locations. A spatial-temporal cloaking check-in model following k-anonymity principle is introduced. And methods to extract two kinds of ROIs, popular regions and personal regions, are proposed respectively. Experimental results illustrate that by analyzing the characteristics of those protected locations, ROIs are able to be discovered as well. Furthermore, our work shows that privacy protection and personalized services can be both achieved in LBSNS.

A Group Based Context-Aware Strategy for Mobile Collaborative Applications

With significant accomplishments of wireless communication, mobile collaborative applications indispensably and inevitably become a new trend. Traditional context aware applications are not directly practicable to current mobile collaborative environments because: 1) Traditional context-aware applications mainly aim at one single user. But in a collaborative environment, the object is a group of users. 2) Traditional context-aware services are always static. In a mobile environment, however, context aware services keep on continuously changing. Fixed communication mode also changes into collaboration in motion. In this paper, a context-aware organization model for mobile collaboration and its management mechanism are proposed. We also introduce a collaborative recommend strategy based on above model. At last, the strategies we proposed are implemented in the Location Aware Mobile Cooperative System (LaMOC) system.

A Hybrid Spatial-Temporal Index for location based service

Nowadays, numerous applications involving spatial data are emerging. While efficient management of tremendous spatial-temporal objects is a challenge due to the highly dynamic nature of the data, the demand for fast getting query results and minimum update cost. However, the query performance optimizations and the update cost minimizations cannot be achieved at the same time. In this paper, a novel approach is presented to get a tradeoff between those two. A spatial-temporal index mechanism named as Hybrid Spatial-Temporal Index (HSTI), which satisfies the time slice, time interval, event, and trajectory queries is introduced. The effectiveness and efficiency of the proposed index is validated through extensive experiments.