Kyungmi Lee

Exploration of geo-tagged photos through data mining approaches

With the development of web technique and social network sites human now can produce information, share with others online easily. Photo-sharing website, Flickr, stores huge number of photos where people upload and share their pictures. This research proposes a framework that is used to extract associative points-of-interest patterns from geo-tagged photos in Queensland, Australia, a popular tourist destination hosting the great Barrier Reef and tropical rain forest. This framework combines two popular data mining techniques: clustering for points-of-interest detection, and association rules mining for associative points-of-interest patterns. We report interesting experimental results and discuss findings.

A generic triangle-based data structure of the complete set of higher order Voronoi diagrams for emergency management

Abstract We introduce a generic Delaunay triangle-based data structure for geoinformation processing in disaster and emergency management. The data structure supports the complete set of higher order Voronoi diagrams (order- k ) Voronoi diagrams, ordered order- k Voronoi diagrams, and k -th nearest Voronoi diagrams for all ( k ) . It provides useful and insightful information for what-if nearest queries, what-if neighboring queries, what-if zoning queries, what-if facility locating queries and what-if routing queries to handle various scenarios in the four stages of emergency management (mitigation, preparedness, response and recovery). We also demonstrate how the complete set of higher order Voronoi diagrams can be used for each phase of emergency management in diverse geoinformatics environments.

Feature extraction and gating techniques for ultrasonic shaft signal classification

Discrete wavelet transform (DWT) coefficients of ultrasonic test signals are considered useful features for input into classifiers due to their effective time-frequency representation of non-stationary signals. However, DWT exhibits a time-variance problem that has resulted in reservations for its wide acceptance. In this paper, a new technique to derive a preprocessing method for time-domain A-scans signal is presented. This technique offers consistent extraction of a segment of the signal from long signals that occur in the non-destructive testing of shafts. Two different classifiers using artificial neural networks and support vector machines are supplied with features generated by our new preprocessing method and their classification performance are compared and evaluated. Their performances are also compared with other alternatives and report the results here. This investigation establishes experimentally that DWT coefficients can be used as a feature extraction scheme more reliably by using our new preprocessing technique.

Mining Points-of-Interest Association Rules from Geo-tagged Photos

The advent of photo-sharing services results in massive user-generated geo-tagged photos. These photos implicitly and explicitly indicate points-of-interest and their associations. This study aims to combine two data mining techniques: clustering and association rules mining to mine areas of attraction, and their associative patterns. We analyze photos from Flickr in the area of Queensland, Australia, a popular tourist destination hosting the Great Barrier Reef and tropical rain forest. We report interesting experimental results and discuss findings.

Sequential pattern mining of geo-tagged photos with an arbitrary regions-of-interest detection method

Propose an arbitrary shape detection regions-of-interest mining algorithm.Propose a sequential trajectory pattern mining framework for Flickr geo-tagged photos.First time analyzing geo-tagged photos using sequential trajectory pattern mining.Experiments supporting the applicability, and effectiveness of our proposed algorithm. Geo-tagged photos leave trails of movement that form trajectories. Regions-of-interest detection identifies interesting hot spots where many trajectories visit and large geo-tagged photos are uploaded. Extraction of exact shapes of regions-of-interest is a key step to understanding these trajectories and mining sequential trajectory patterns. This article introduces an efficient and effective grid-based regions-of-interest detection method that is linear to the number of grid cells, and is able to detect arbitrary shapes of regions-of-interest. The proposed algorithm is combined with sequential pattern mining to reveal sequential trajectory patterns. Experimental results reveal quality regions-of-interest and promising sequential trajectory patterns that demonstrate the benefits of our algorithm.

Fast action recognition using negative space features

Due to the number of potential applications and their inherent complexity, automatic capture and analysis of actions have become an active research area. In this paper, an implicit method for recognizing actions in a video is proposed. Existing implicit methods work on the regions of subjects, but our proposed system works on the surrounding regions, called negative spaces, of the subjects. Extracting features from negative spaces facilitates the system to extract simple, yet effective features for describing actions. These negative-space based features are robust to deformed actions, such as complex boundary variations, partial occlusions, non-rigid deformations and small shadows. Unlike other implicit methods, our method does not require dimensionality reduction, thereby significantly improving the processing time. Further, we propose a new method to detect cycles of different actions automatically. In the proposed system, first, the input image sequence is background segmented and shadows are eliminated from the segmented images. Next, motion based features are computed for the sequence. Then, the negative space based description of each pose is obtained and the action descriptor is formed by combining the pose descriptors. Nearest Neighbor classifier is applied to recognize the action of the input sequence. The proposed system was evaluated on both publically available action datasets and a new fish action dataset for comparison, and showed improvement in both its accuracy and processing time. Moreover, the proposed system showed very good accuracy for corrupted image sequences, particularly in the case of noisy segmentation, and lower frame rate. Further, it has achieved highest accuracy with lowest processing time compared with the state-of-art methods.

Mining Frequent Trajectory Patterns and Regions-of-Interest from Flickr Photos

Flickr represents a massive opportunity to mine valuable human movement data from geo-tagged photos. However, existing Flickr trajectory data mining research has not considered mining frequent trajectory patterns whilst also considering the temporal domain. Therefore, a significant opportunity exists to demonstrate the application of a pattern mining algorithm to a large geo-tagged photo dataset. Thus, we present a novel application of the trajectory pattern mining algorithm to a 2012 Flickr dataset of Australia and encompassing state, Queens land. In our experiments we show that many interesting, previously unknown patterns discovered through our framework. Our framework is able to discover expected major landmarks such as cities and tourist attractions. In addition, we make the notable discover of what is theorized to be valuable tourist travel information about sequential movements between hot-spot attractions.

Points-of-Interest Mining from People's Photo-Taking Behavior

Millions of geo-tagged photos are becoming available due to the widespread of photo-sharing websites. These social medias capture attractive points-of-interest and contain interesting photo-taking patterns. Massive amount of these user-oriented data produces new challenges and understanding peoples photo-taking behavior is of great importance for local tourism-related businesses. This paper analyzes geotagged photos from Flickr for Queensland, a tourismintensive and the second largest state in Australia. We report interesting points-of-interest patterns and discuss these findings.

Map segmentation for geospatial data mining through generalized higher-order Voronoi diagrams with sequential scan algorithms

Segmentation is one popular method for geospatial data mining. We propose efficient and effective sequential-scan algorithms for higher-order Voronoi diagram districting. We extend the distance transform algorithm to include complex primitives (point, line, and area), Minkowski metrics, different weights and obstacles for higher-order Voronoi diagrams. The algorithm implementation is explained along with efficiencies and error. Finally, a case study based on trade area modeling is described to demonstrate the advantages of our proposed algorithms.

Mining qualitative patterns in spatial cluster analysis

Clustering is an important concept formation process within AI. It detects a set of objects with similar characteristics. These similar aggregated objects represent interesting concepts and categories. As clustering becomes more mature, post-clustering activities that reason about clusters need a great attention. Numerical quantitative information about clusters is not as intuitive as qualitative one for human analysis, and there is a great demand for an intelligent qualitative cluster reasoning technique in data-rich environments. This article introduces a qualitative cluster reasoning framework that reasons about clusters. Experimental results demonstrate that our proposed qualitative cluster reasoning reveals interesting cluster structures and rich cluster relations.