Skip to main content

Über dieses Buch

This book constitutes the refereed proceedings of the 8th International Conference on Active Media Technology, AMT 2012, held in Macau, China, in December 2012. The 65 revised full papers were carefully reviewed and selected from a numerous submissions. The papers are organized in topical sections on awareness multi-agent systems, data mining, ontology mining, web reasoning, social applications of active media, human-centered computing, personalization and adaptation, smart digital art and e-learning.



Active Media Data Mining and Machine Learning Techniques

Movie Genre Classification Using SVM with Audio and Video Features

In this paper, we propose a movie genre classification system using a meta-heuristic optimization algorithm called Self-Adaptive Harmony Search (i.e., SAHS) to select local features for corresponding movie genres. Then, each one-against-one Support Vector Machine (i.e., SVM) classifier is fed with the corresponding local feature set and the majority voting method is used to determine the prediction of each movie. Totally, we extract 277 features from each movie trailer, including visual and audio features. However, no more than 25 features are used to discriminate each pair of movie genres. The experimental results show that the overall accuracy reaches 91.9%, and this demonstrates more precise features can be selected for each pair of genres to get better classification results.

Yin-Fu Huang, Shih-Hao Wang

Automatic Player Behavior Analyses from Baseball Broadcast Videos

In this paper, we present a baseball player behavior analysis system by combining pitch types and swing events. We use eight kinds of semantic scenes detected from baseball videos in our previous work. For the pitch types, we use the characteristic of the ball in a pitch scene to identify the ball trajectory, and then 39 features are extracted to feed into a trained SVM for classifying pitch types. For the swing events, we use moving objects in the batter region to determine whether a swing occurs. Then, the event following the swing is detected using an HMM, based on the after-swing scene sequence. Next, the experimental results show that both pitch type recognition and swing event detection have accuracy rates 91.5% and 91.1%. Finally, we analyze and summarize player behavior by combining pitch types and swing events.

Yin-Fu Huang, Zong-Xian Yang

Hot Topic Detection in News Blogs from the Perspective of W2T

News blog hot topics are important for the information recommendation service and marketing. However, information overload and personalized management make the information arrangement more difficult. Moreover, what influences the formation and development of blog hot topics is seldom paid attention to. In order to correctly detect news blog hot topics, the paper first analyzes the development of topics in a new perspective based on W2T (Wisdom Web of Things) methodology. Namely, the characteristics of blog users, context of topic propagation and information granularity are unified to analyze the related problems. Some factors such as the user behavior pattern, network opinion and opinion leader are subsequently identified to be important for the development of topics. Then the topic model based on the view of event reports is constructed. At last, hot topics are identified by the duration, topic novelty, degree of topic growth and degree of user attention. The experimental results show that the proposed method is feasible and effective.

Erzhong Zhou, Ning Zhong, Yuefeng Li, Jia-jin Huang

A Clustering Ensemble Based on a Modified Normalized Mutual Information Metric

It has been proved that ensemble learning is a solid approach to reach more accurate, stable, robust, and novel results in all data mining tasks such as clustering, classification, regression and etc. Clustering ensemble as a sub-field of ensemble learning is a general approach to improve the performance of clustering task. In this paper by defining a new criterion for clusters validation named Modified Normalized Mutual Information (MNMI), a clustering ensemble framework is proposed. In the framework first a large number of clusters are prepared and then some of them are selected for the final ensemble. The clusters which satisfy a threshold of the proposed metric are selected to participate in final clustering ensemble. For combining the chosen clusters, a co-association based consensus function is applied. Since the Evidence Accumulation Clustering (EAC) method can’t derive the co-association matrix from a subset of clusters, Extended Evidence Accumulation Clustering (EEAC), is applied for constructing the co-association matrix from the subset of clusters. Employing this new cluster validation criterion, the obtained ensemble is evaluated on some well-known and standard datasets. The empirical studies show promising results for the ensemble obtained using the proposed criterion comparing with the ensemble obtained using the standard clusters validation criterion.

Hamid Parvin, Behzad Maleki, Sajad Parvin

Influence of Erroneous Pairwise Constraints in Semi-supervised Clustering

Side information such as pairwise constraints is useful to improve the clustering performance in general. However, constraints are not always error free in general. When erroneous constraints are specified as side information, treating them as hard constraints could have the disadvantage since strengthening incorrect or erroneous constraints can lead to performance degradation. In this paper we conduct extensive experiments to investigate the influence of erroneous pairwise constraints over various document datasets. Several state-of-the-art semi-supervised clustering methods with graph representation were evaluated with respect to the type of constraints as well as the number of constraints. Experimental results confirmed that treating pairwise constraints as hard constraints is vulnerable to erroneous ones. However, the results also revealed that the influence of erroneous constraints depends on how the constraints are exploited inside a learning algorithm.

Tetsuya Yoshida

User Correlation Discovery and Dynamical Profiling Based on Social Streams

In this study, we try to discover the potential and dynamical user correlations using those reorganized social streams in accordance with users’ current interests and needs, in order to assist the information seeking process. We develop a mechanism to build a Dynamical Socialized User Networking (DSUN) model, and define a set of measures (such as interest degree, and popularity degree) and concepts (such as complementary tie, weak tie, and strong tie), which can discover and represent users’ current profiling and dynamical correlations. The corresponding algorithms are developed respectively. Based on these, we finally discuss an application scenario of the DSUN model with experiment results.

Xiaokang Zhou, Qun Jin

Extraction of Human Social Behavior from Mobile Phone Sensing

With lots of sensors built in, mobile phones become a pervasive platform for seamlessly sensing of human behaviors. In this paper, we investigate how to use location data and communication records collected from mobile phones to obtain human social interaction features and activity patterns. Social Interaction features refer to the temporal and spatial interactive information, and activity patterns include movement patterns. Meanwhile, the similarities and differences of human behaviors at different ages, as well as distinct occupations are analyzed. The results indicate that different population has a diversity of social interaction and activity patterns, and human social behaviors are highly associated with age and occupation. Furthermore, we make a correlation analysis about social temporal interaction, social spatial interaction and social activity, which lead us to conclude that the three elements are interrelated among young people but not middle-ages. Our work could be a cornerstone for research of personalized psychological health assistance based on mobile phone data.

Minshu Li, Haipeng Wang, Bin Guo, Zhiwen Yu

Continuity of Defuzzification on L2 Space for Optimization of Fuzzy Control

The purpose of this study is to consider the fuzzy optimal control based on the functional analysis. We used a mathematical approach to compute optimal solutions. The feedback of fuzzy control is evaluated through approximate reasoning using the center of sums defuzzification method or the height method on IF-THEN fuzzy rules. The framework consists of two propositions: To guarantee the convergence of optimal solution, a set of fuzzy membership functions (admissible fuzzy controller) which are selected out of continuous function space is compact metrizable. And assuming approximate reasoning to be a functional on the set of membership functions, its continuity is proved. Then, we show the existence of a fuzzy controller which minimizes (maximizes) the integral performance function of the nonlinear feedback fuzzy system.

Takashi Mitsuishi, Takanori Terashima, Nami Shimada, Toshimichi Homma, Kiyoshi Sawada, Yasunari Shidama

A On-Line News Documents Clustering Method

To improve the efficiency and accuracy of on-line news event detection (ONED) method, we select the words that their term frequency (TF) is greater than a threshold to create the vector space model of the news document, and propose a two-stage clustering method for ONED. This method divides the detection process into two stages. In the first stage, the similar documents collected in a certain period of time are clustered into micro-clusters. In the second stage, the micro-clusters are compared with previous event clusters. The experimental results show that the proposed method has fewer computation load, higher computing rate, and less loss of accuracy.

Hui Zhang, Guo-hui Li, Xin-wen Xu

Agent-Based Applications and Multi-agent Systems

Distributed Protocols for Multi-Agent Coalition Formation: A Negotiation Perspective

We investigate collaborative multi-agent systems and how they can use simple, scalable negotiation protocols to coordinate in a fully decentralized manner. Specifically, we study multi-agent

distributed coalition formation

. We summarize our past and ongoing research on collaborative coalition formation and describe our original distributed coalition formation algorithm. The present paper focuses on negotiation-based view of coalition formation in collaborative

Multi-Agent Systems

(MAS). While negotiation protocols have been extensively studied in the context of competitive, self-interested agents, we argue that negotiation-based approach may be potentially very useful in the context of collaborative agents, as well – as long as those agents, due to limitations of their sensing and communication abilities, have different views of and preferences over the states of the world. In particular, we show that our coalition formation algorithm has several important, highly desirable properties when viewed as a negotiation protocol.

Predrag T. Tošić, Carlos Ordonez

Multi-Agent Liquidity Risk Management in an Interbank Net Settlement System

A net settlement system is a payment system between banks, where a large number of transactions are accumulated, usually waiting until the end of each day to be settled through payment instruments like: wire transfers, direct debits, cheques, .... These systems also provide clearing functions to reduce interbank payments but are sometimes exposed to liquidity risks. Monitoring, and optimizing the interbank exchanges through suitable tools is useful for the proper functioning of these systems. The goal is to add to these systems an intelligent software layer integrated with the existing system for the improvement of transactions processing and consequently avoid deadlock situations, deficiencies and improve system efficiency. We model and develop by multi-agent an intelligent tracking system of the interbank exchanged transactions to optimize payments settlement and minimize liquidity risks.

Badiâa Hedjazi, Mohamed Ahmed-Nacer, Samir Aknine, Karima Benatchba

An Enhanced Mechanism for Agent Capability Reuse

Existing research work on intelligent agents is facing the challenge of extending agent capabilities. One of major difficulties of applying intelligent agents in various software platforms lies in that generalized agent design models may not fit with specified organizational applications. This problem dramatically hampers the practical applications of intelligent agents. In this paper we point out several aspects that might affect the development of agent-based technology; and further address the information reuse and unification issues through suggesting an enhanced agent capability reuse mechanism, which is able to provide an efficient process for agent capability reuse. The proposed design can fulfill the needs for a dynamic agent environment.

Hao Lan Zhang, Jiming Liu, Chaoyi Pang, Xingsen Li

A Bayesian Network Approach to Investigating User-Robot Personality Matching

Personality analysis has been an important topic in both psychology and human-robot interaction (HRI). The main theme of this paper is to explore the relationship between individuals’ personality traits and their tactile interaction patterns with a robot. A sociable robot, Pleo, was used in the experiment. The tactile interaction patterns of the participants with the robot were video-recorded and analyzed. Bayesian network (BN) classifiers such as NBN (naïve BN), TAN (tree-augmented BN), and GBN (general BN) were used to examine the causal relationship between personality traits and touch patterns. The analysis showed that individuals’ personality traits could be inferred based on their tactile interaction patterns with a robot. What-if and goal-seeking analysis using GBN confirmed this result. The findings of this paper are promising and its implications are discussed.

Jungsik Hwang, Kun Chang Lee, Jaeyeol Jeong

Modelling Multi-Criteria Decision Making Ability of Agents in Agent-Based Rice Pest Risk Assessment Model

This paper aims to introduce an agent-based multi-criteria assessment model that support stakeholders to make multi-criteria decisions for rice pest management, this study designs agents with dynamic ability of multi-criteria decision making from information of other agents so that rice pest risk maps are built. A case study is carried out on risk assessment of brown plant hoppers in the Mekong Delta region, simulation results show that rice pest risk index aggregated from criteria and agent-based models are useful tools to support rice pest management.

Vinh Gia Nhi Nguyen, Hiep Xuan Huynh, Alexis Drogoul

Agent Based Assistance for Electric Vehicles An Evaluation

Even before car manufacturers start offering series-produced electric vehicles in a large scale, expectations in the electric powertrain are considerably high. Prospective business perspectives are additionally driven by the so called


technology, which allows electric vehicles to not only procure electric energy, but also to feed energy back into the grid network. However, by using Vehicle-to-Grid, energy literally degenerates into an article of merchandise and becomes of interest to several stakeholders. We have developed a multi-agent system, which embraces this exact view and maximises the interest of several stakeholders in using Vehicle-to-Grid capable electric vehicles. The purpose of this paper is to describe the evaluation of our assistance system and to present collected evaluation results.

Marco Lützenberger, Jan Keiser, Nils Masuch, Sahin Albayrak

Ubiquitous Intelligent Devices and Systems

Event Calculus-Based Adaptive Services Composition Policy for AmI Systems

Services composition technology which is used in Ambient Intelligence should have the features of context-aware, partial order and concurrent. It should adaptive to the dynamic change of user preference and context. To meet these requirements, the paper puts forward an adaptive dynamic services composition framework and its implementation mechanism based on event calculus. It studies the basic principles and technologies for descripting domain services, context information and domain rules based on event calculus. On the basis, it details the composition planning mechanism. At last, a prototype system, intelligence application control, is built to verify the effectiveness and availability of the services composition policy.

Huibing Zhang, Jingwei Zhang, Ya Zhou, Junyan Qian

QoS- and Resource-Aware Service Composition and Adaptation

Flexible and adaptive quality-of-service (QoS) is desirable for distributed real-time applications, such as e-commerce, or multimedia applications. The objective of this research is to dynamically instantiate composite services by effectively utilising the collective capabilities of the resources to deliver distributed applications. Related to this objective are the problems of: (1) predicting system and network resources utilisation as well as the user’s changing requirements on the provided services, and (2) finding optimal execution plans for a service that meet end-to-end quality requirements and adapting the available resources in accordance to the changing situation. This paper presents a framework for adaptive QoS and resource management in provisioning composite services. We also develop distributed algorithms for finding the multi-constrained optimal execution plan to enable delivery of QoS-assured composite services.

Quoc Bao Vo, Minyi Li

An Event-Driven Energy Efficient Framework for Wearable Health-Monitoring System

Wearable health-monitoring system requires keeping the balance between energy consuming and user’s real-time service demands with continuous sensing, wireless communication and processing. In this paper, we present a design framework for an Energy Efficient Wearable Health-monitoring System (EEWHMS). EEWHMS uses smartphone as a central unit to process data from wearable sensors, with event-driven energy management strategy to save energy. State transitions and continuous being still of user is used to adjust duty cycle of the system. We present the design, implementation, and evaluation of EEWHMS, which collects user’s information with accelerometer and physiological sensors, and sends it to an Android phone by Bluetooth. According to evaluation of power of Bluetooth chip, CPU load of smartphone and response time when emergency happened, the system demonstrates its capability to keep balance between real time and long term sensing in an energy-efficient manner.

Na Li, YiBin Hou, ZhangQin Huang

Learning Style Model for e-Learning Systems

E-learning is the use of digital devices and networks to transfer skills and knowledge to learners. E-learning applications and tools include active media, multimedia, web-based learning, computer-based learning, virtual education tools, etc. In such rich learning environment learners need a method to specify a clear learning path that matches with their learning preferences. Learning style model can provide a basis for such methods. In this paper we study and implement a web-based learning style model that can help learners to find a learning path in an e-learning system.

Mohamed Hamada

A Trajectory-Based Recommender System for Tourism

Recommendation systems provide focused information to users on a set of objects belonging to a specific domain. The proposed recommender system provides personalized suggestions about touristic points of interest. The system generates recommendations, consisting of touristic places, according to the current position of a tourist and previously collected data describing tourist movements in a touristic location/city. The touristic sites correspond to a set of points of interest identified a priori. We propose several metrics to evaluate both the spatial coverage of the dataset and the quality of recommendations produced. We assess our system on two datasets: a real and a synthetic one. Results show that our solution is a viable one.

Ranieri Baraglia, Claudio Frattari, Cristina Ioana Muntean, Franco Maria Nardini, Fabrizio Silvestri

An Adaptive Method for the Tag-Rating-Based Recommender System

In this paper, we propose an adaptive method for recommender system based on users’ preference to items represented by the ratings of users. This method defines a term-association matrix to describe the relation between tags and items properties. A gradient descent method is employed to compute the association matrix. The association matrix is then used to implement the two kinds of recommendation, namely, tag recommendation and items properties recommendation.

Xi Yuan, Jia-jin Huang

Comparative Study of Joint Decision-Making on Two Visual Cognition Systems Using Combinatorial Fusion

In processing multimedia technologies or decision-making in visual cognition systems, combination by both simple average and weighted average are used. In this paper, we extend each visual cognition system to a scoring system using Combinatorial Fusion Analysis (CFA). We investigate the performance of the combined system in terms of individual system’s performance and confidence. Twelve experiments are conducted and our main results are: (a) The combined systems perform better only if the two individual systems are relatively good, and (b) overall, rank combination is better than score combination. In addition, we compare the three types of averages: simple average M


, weighted average M




, and weighted average M





, where


is related to confidence of each system. Our results exhibit a novel way to better make joint decisions in visual cognition using Combinatorial Fusion.

Amy Batallones, Cameron McMunn-Coffran, Kilby Sanchez, Brian Mott, D. Frank Hsu

Active Media Based Information Retrieval and Processing

Research on Touch as a Means of Interaction in Digital Art

This paper explores touch as a medium of creative expression in digital art. Important to this work is haptic interaction which takes place via a physical interface. Touch is not simply a tactile sensation, but also a way to explore an environment by combining sensory information such as pressure and temperature. Therefore an artist’s creation can be understood by the viewer through the sensory information received. The main focus of the paper is on the sense of touch and it will help our understanding to investigate the following four artworks and corresponding technologies. The selected works include both visual and audio interactions.

Touch Me

concentrates on the sensory experience perceived through the skin, while

A-Volve, Colorful Touch Palette


FuSA2 Touch Display

are concerned with the virtual hand in technological systems. I believe that multi-touch, which enables a user to interact with more than one finger, will become the most important tool for experiencing sensory digital art in the future.

Mingwei Zhang, Guoyong Dai

Perceptual Image Hashing with Histogram of Color Vector Angles

Image hashing is an emerging technology for the need of, such as image authentication, digital watermarking, image copy detection and image indexing in multimedia processing, which derives a content-based compact representation, called image hash, from an input image. In this paper we study a robust image hashing algorithm with histogram of color vector angles. Specifically, the input image is first converted to a normalized image by interpolation and low-pass filtering. Color vector angles are then calculated. Thirdly, the histogram is extracted for those angles in the inscribed circle of the normalized image. Finally, the histogram is compressed to form a compact hash. We conduct experiments for evaluating the proposed hashing, and show that the proposed hashing is robust against normal digital operations, such as JPEG compression, watermarking embedding, scaling, rotation, brightness adjustment, contrast adjustment, gamma correction, and Gaussian low-pass filtering. Receiver operating characteristics (ROC) curve comparisons indicate that our hashing performs much better than three representative methods in classification between perceptual robustness and discriminative capability.

Zhenjun Tang, Yumin Dai, Xianquan Zhang, Shichao Zhang

Data Hiding Method Based on Local Image Features

We propose a data hiding method with high embedding capacity and good fidelity. It is done by block classification, data embedding, and pixel adjustment. The image blocks are firstly divided into three categories: smooth block, edge block and textural block. Different secret bits are then adaptively embedded into different blocks in terms of the image types by using least-significant-bit (LSB) substitution. After data embedding, the changed pixels are adjusted to minimize distortion. Many experiments are conducted to validate the effectiveness of the proposed method.

Xianquan Zhang, Zhenjun Tang, Tao Liang, Shichao Zhang, Yingjun Zhu, Yonghai Sun

Fast Flow Visualization on CUDA Based on Texture Optimization

Flow visualization plays an important role in many scientific visualization applications. It is effective to visualize flow fields with moving textures which vividly capture the properties of flow field through varying texture appearances.Texture-optimization-based (TOB) flow visuliaztion can produce excellent visualization results of flow fields. However, TOB flow visualization without acceleration is time-consuming. In this paper, we propose fast flow visualization based on the accelerated parallel TOB flow visualization which is entirely implemented on CUDA. High performance is achieved since most time-consuming computations are performed in parallel on GPU and data transmission between CPU and GPU are arranged properly. The experimental results show that our TOB flow visualization generates results with fast synthesis speed and high synthesis quality.

Ying Tang, Zhan Zhou, Xiao-Ying Shi, Jing Fan

A Message Passing Graph Match Algorithm Based on a Generative Graphical Model

In This paper, we present a generative model to measure the graph similarity, assuming that an observed graph is generated from a template by a Markov random field. The potentials of this random process are characterized by two sets of parameters: the attribute expectations specified by the the template graph, and the variances that can be learned by a maximum likelihood estimator from a collection of samples. Once a sample graph is observed, a max-product loopy belief propagation algorithm is applied to approximate the most probable explanation of the template’s vertices, mapped to the sample’s vertices. As demonstrated by the experiments, compared with other algorithms, the proposed approach performed better for near isomorphic graphs in the typical graph alignment and information retrieval applications.

Gang Shen, Wei Li

Fast Content-Based Retrieval from Online Photo Sharing Sites

Literally billions of images have been uploaded to photo sharing sites since their inception, comprising a staggering wealth of visual information. However, effective tools for querying these collections are rare and keyword based. Since users rarely annotate their images, this approach is only of limited use. Content-based image retrieval (CBIR) extracts features directly from images and bases searches on these features. However, conventional CBIR approaches require a dedicated system that performs feature extraction during photo upload and a database system to store the features, and are hence not available to the average user. In this paper, we present a very fast content-based retrieval method that performs feature extraction on-the-fly during the retrieval process and thus can be employed client-side on images downloaded from photo sharing sites such as Flickr.

Our approach is based on the fact that images uploaded to Flickr are stored in a JPEG format optimised to minimise disk space and bandwidth usage. In particular, we exploit the optimised Huffman compression tables, which are stored in the JPEG headers, as image descriptors. Since, in contrast to other approaches, we thus have to read only a fraction of the image file and similarity calculation is of low complexity, our approach is extremely fast as demonstrated by the bandwidth used to retrieve images from the Flickr photo sharing site. We also show that nevertheless retrieval performance is comparable to CBIR using colour histograms which is at the core of many CBIR systems.

Gerald Schaefer, David Edmundson

Interactive Exploration of Image Collections on Mobile Devices

Image collections are ever growing and hence visual information is becoming more and more important. Moreover, the classical paradigm of taking pictures has changed, first with the spread of digital cameras and, more recently, with mobile devices equipped with integrated cameras. Clearly, these image repositories need to be managed, and tools for effectively and efficiently searching image databases are highly sought after, especially on mobile devices where more and more images are being stored. In this paper, we present an image browsing system for interactive exploration of image collections on mobile devices. Images are arranged so that visually similar images are grouped together while large image repositories become accessible through a hierarchical, browsable tree structure, arranged on a hexagonal lattice. The developed system provides an intuitive and fast interface for navigating through image databases using a variety of touch gestures.

Gerald Schaefer, Matthew Tallyn, Daniel Felton, David Edmundson, William Plant

The Derived Kernel Based Recognition Method of Vehicle Type

This paper applies the updated derived kernel algorithm into the vehicle type recognition, which is a heated research area based on the method of pattern recognition and digital image processing because of its significant usage on exit and entry administration, traffic and vehicle control and toll collection. The method of two- layer derived kernel on neural response is involved in extracting useful features from vehicle images, for the method itself has better capacity of decreasing the negative influences from different colors and vehicle speeds, background condition interference and blur noises. Some clustering algorithms are employed on the process of templates construction, and the first nearest neighbor algorithm on pattern classification. Since our method can get rid of the disturbances from similar parts of vehicle images and make the most of the features of representative parts, the vehicle type recognition accuracy reaches above 95% as high.

Zhen Chao Zhang, Yuan Yan Tang, Chang Song Liu, Xiao Qing Ding

An Approach to Define Flexible Structural Constraints in XQuery

This paper presents a formal definition of an extension of the XQuery Full-Text language: the proposed extension consists in adding two new flexible axes, named




, which express structural constraints that can be specified by the user. Both constraints are evaluated in an approximate way with respect to a considered path, and their evaluation produces a path relevance score for each retrieved element. The formal syntax and the semantics of the two new axis are presented and discussed.

Emanuele Panzeri, Gabriella Pasi

DC Stream Based JPEG Compressed Domain Image Retrieval

The vast majority of images are stored in compressed JPEG format. When performing content-based image retrieval, faster feature extraction is possible when calculating them directly in the compressed domain, avoiding full decompression of the images. Algorithms that operate in this way calculate image features based on DCT coefficients and hence still require partial decoding of the image to arrive at these. In this paper, we introduce a JPEG compressed domain retrieval algorithm that is based not directly on DCT coefficients but on differences of these, which are readily available in a JPEG compression stream. In particular, we utilise solely the DC stream of JPEG files and make direct use of the fact that DC terms are differentially coded. We build histograms of these differences and utilise them as image features, thus eliminating the need to undo the differential coding as in other methods. In combination with a colour histogram, also extracted from DC data, we show our approach to give (to our knowledge) the best retrieval accuracy of a JPEG domain retrieval algorithm, outperforming other compressed domain methods and reaching a performance close to that of the best performing MPEG-7 descriptor.

Gerald Schaefer, David Edmundson

Semantic Computing for ActiveMedia, Social Networks, and AMT-Based Systems

A Comparative Study of Community Structure Based Node Scores for Network Immunization

Network immunization has often been conducted by removing nodes with large network centrality so that the whole network can be fragmented into smaller subgraphs. Since contamination (e.g., virus) is propagated among subgraphs (communities) along links in a network, besides centrality, utilization of community structure seems effective for immunization. We have proposed community structure based node scores in terms of a vector representation of nodes in a network. In this paper we report a comparative study of our node scores over both synthetic and real-world networks. The characteristics of the node scores are clarified through the visualization of networks. Extensive experiments are conducted to compare the node scores with other centrality based immunization strategies. The results are encouraging and indicate that the node scores are promising.

Yuu Yamada, Tetsuya Yoshida

Semantic Precision and Recall for Evaluating Incoherent Ontology Mappings

Ontology mapping plays an important role in the Semantic Web, which generates correspondences between different ontologies. Usually, precision and recall are used to evaluate the performance of a mapping method. However, they do not take into account of the semantics of the mapping. Thus, semantic precision and recall are proposed to resolve the restricted set-theoretic foundation of precision and recall. But the semantic measures do not consider the incoherence in a mapping which causes some trivialization problems. In this paper, we propose semantic measures for evaluating

incoherent ontology mappings

. Specifically, a general definition of semantic measures is given based on a set of formal definitions capturing reasoning with incoherent mappings. Then we develop a concrete approach to reasoning with incoherent mappings, which results in some specific semantic measures. Finally, we conduct experiments on the data set of conference track provided by OAEI.

Qiu Ji, Zhiqiang Gao, Zhisheng Huang, Man Zhu

A Self-organization Method for Reorganizing Resources in a Distributed Network

Using bio-inspired agents to reorganize resource has been adopted to address the distributed resource optimization issue in distributed networks. This paper presents a self-organization algorithm to reorganize resources by the use of autonomous agents which can exchange their resources each other. Agents are equipped with two operations (pull and push operation) and three behaviors (best selection, better selection and random selection). And at every moment, agents probabilistically choose a behavior to perform. Experimental results indicate that the strategy has a positive influence on system performance.

Xiaolong Guo, Jia-jin Huang

Extracting Property Semantics from Japanese Wikipedia

Here is discussed how to build up ontology with many properties from Japanese Wikipedia. The ontology includes is-a relationship (rdfs:subClassOf), class-instance relationship (rdf:type) and synonym relation (skos:altLabel) moreover it includes property relations and types. Property relations are triples, property domain (rdfs:domain) and property range (rdfs:range). Property types are object (owl:ObjectProperty), data (owl:DatatypeProperty), symmetric (owl:SymmetricProperty), transitive (owl:TransitiveProperty), functional (owl:FunctionalProperty) and inverse functional (owl:InverseFunctionalProperty).

Susumu Tamagawa, Takeshi Morita, Takahira Yamaguchi

An Ontology Based Privacy Protection Model for Third-Party Platform

A third-party trading platform is a web based system that provides services for sellers and buyers. On such platform, users are required to provide personal information to ensure the authenticity and undeniability of a transaction. In this paper, we propose an ontology based privacy protection model for third-party platform, which allows buyers and sellers to define privacy policies according to their preferences and converts policies into ontology based forms. We introduce the property of


sellers who require the minimal information from buyers while satisfying other trading requirements. The proposed policy matching algorithm finds such sellers as candidates for a buyer request. A practical example is given to illustrate our model.

Haojun Yu, Yuqing Sun, Jinyan Hu

Evaluating Ontology-Based User Profiles

User profiles play an important role in any process of personalization as they represent the user’s interests and preferences. Only if a user profile faithfully represents the information related to a user a system may rely on it. This paper shortly presents a comparative evaluation between two distinct approaches that analyze textual documents for defining user profiles based on the usage of the YAGO general purpose ontology. The performed evaluations compare the two approaches both by the robust index measure and their efficiency.

Silvia Calegari, Gabriella Pasi

Semantic Information with Type Theory of Acyclic Recursion

The paper provides introduction into the language, semantics, and the reduction calculus of acyclic recursion. The expressiveness of the theory for conveying semantic information, relevant for applications in active media technologies, is demonstrated by representing naming information.

Roussanka Loukanova

Pyxis: An Active Replication Approach for Enhancing Social Media Services

To support the rapid increase of social media, social media services are demanded to be efficient, highly available and scalable. The paper provides an active replication approach Pyxis to enhance social media services. Following Pyxis, either the newly-built social media services or the rebuilt ones can tolerate site failures and network partition failures, and thus achieve better user experience by providing continuous services on the basis of keeping required causal consistency. The experimental results show that Pyxis can adapt to a large number of concurrent users, and recover from failures in acceptable intervals.

Sen Li, Beihong Jin, Yuwei Yang, Wenjing Fang

Social Network Analysis of Virtual Worlds

As 3D environments become both more prevalent and more fragmented, studying how users are connected via their avatars and how they benefit from the virtual world community has become a significant area of research. An in depth analysis of the virtual world social networks is necessary to evaluate its worlds, to understand the impact of avatar social networks on the virtual worlds, and to improve future online social networks. Our current efforts are focused on building and exploring the social network aspects of virtual worlds. In this paper we evaluate the Second Life social network we have created and compare it to other social networking sites found on the web. Experimental results with data crawled from Second Life virtual worlds demonstrate that our approach was able to build a representative network of avatars in virtual world from the sample data. The analysis comparison between virtual world social networks and others in flat web allows us to gauge measures that better explore the relationship between locations linked by multiple users and their avatars. Using this comparison, we can also determine if techniques of personalization search and content recommendation are feasible for virtual world environments.

Gregory Stafford, Hiep Luong, John Gauch, Susan Gauch, Joshua Eno

Active Media Framework for Network Processing Components

This paper presents a framework for building active media content and defining network processing for it. It makes two novel contributions. The first introduces network processing as first-class objects like active media content in the sense that components are introduced as the only constituent of our network processing for components as well as active content. It enablesan active media content to be composed of one or more active media or network processing components and to migrate between these components, which may be running on different computers. It also offered several basic operations for network processing, e.g., carrying, forwarding, duplication, and synchronization. The operations can be treated as active media components; they can be dynamically deployed at local or remote computers through GUI-based manipulations. It therefore allows an end-user to easily and rapidly configure network processing in the same way as if he/she had edited the documents. We constructed a prototype implementation of this infrastructure and its applications.

Ichiro Satoh

Semantic Network Monitoring and Control over Heterogeneous Network Models and Protocols,

To accommodate the proliferation of heterogeneous network models and protocols we propose the use of semantic technologies to enable an abstract treatment of networks. Network adapters are employed to lift network specific data into a semantic representation that is grounded in an upper level “NetCore” ontology. Semantic reasoning integrates the disparate network models and protocols into a common RDF-based data model that network applications can be written against without requiring intimate knowledge of the various low level-network details. The system permits the automatic discovery of new devices, the monitoring of device state and the invocation of device actions in a generic fashion that works across network types, including non-telecommunication networks such as social networks. A prototype system called SNoMAC is described that employs the proposed approach operating over UPnP, TR-069 and SIXTH network models and protocols. A major benefit of this approach is that the addition of new models/protocols requires relatively little effort and merely involves the development of a new network adapter based on an ontology grounded in NetCore.

Christopher J. Matheus, Aidan Boran, Dominic Carr, Rem Collier, Barnard Kroon, Olga Murdoch, Gregory M. P. O’Hare, Michael O’Grady

International Workshop on Meta-synthesis and Complex Systems

Opinion Dynamics on Triad Scale Free Network

In this paper, we investigate the opinion dynamics model of social impact theory on triad scale free network with power law degree distribution and tunable clustering coefficient. Based on this opinion dynamic model, we try to observe the clustering coefficient influence on opinion formation by adjusting the triad formation parameter. Simulation result shows that by adjusting triad scale free network parameters, a large clustering coefficient favors development of a consensus. In particular, when in the system with initial opinion proportion of +1, p


=0.5, a consensus seems to be never reached for triad scale free network with any clustering coefficient.

Li Qianqian, Liu Yijun, Tian Ruya, Ma Ning

Distribution of Node Characteristics in Complex Networks of Tree Class

Based on the work of Park-Barabási (PB) we research in detail the (D,H)-phase diagram which describes the correlation and interplay among nodes of complex systems. To do this, we provide a frame of mathematical description, it includes: carrying out symbolization to the assortment of nodes, obtaining symbolic assertive matrix and enumeration formula. Applying the frame to two kinds of tree graphs we find that there exists vivid self-similar motif in the core domain of (D,H)-phase diagram. In order to draw the phase boundary we use a mixed curve of both Cassini oval and ellipse. The stationary of (D,H)-phase diagram is confirmed, but we also have seen a trend that the phase boundary has a phenomenon of little compression when the size of system increases. Finally, we suggest a new classification method to decide dyadic configuration of (D,H)-phase diagram and put it to use in the tree systems.

Ying Tan, Hong Luo, Shou-Li Peng

Dynamic Mergers Drive Industrial Competition Evolution: A Network Analysis Perspective

This paper presents a novel method to explore the relationship between dynamics mergers and the evolution of industrial competition by introducing complex network tool. By taking China’s beer industry as an example, we established Markets-Firms bipartite time series networks, weighted Markets-Firms time series networks and industrial competition time series networks by using the data from 1992 to 2009. Through analyzing the changes of topology index of these networks, we find that dynamic mergers play a key important role in the evolution of industrial competition. The results show that dynamic mergers promote the fragmented local markets to be consolidated into a global market for China’s beer industry, and it also shows the process that competitive relationship among the rivals turns from some single segmented markets to a global cross-market. This paper gives a new view to observe the changing of industrial competition driven by dynamic mergers.

Rui Hou, Jianmei Yang, Canzhong Yao

A Study of Collective Action Threshold Model Based on Utility and Psychological Theories

In this paper, we extend Granovetter’s classic threshold model by adding both utility and psychological threshold. We conduct simulations with the presented model while also considering the spatial factor and friendship influence strength. We observe that the equilibrium of collective dynamics is not closely related to the friendship impact. With no utility and psychological threshold, the equilibrium state of the model is sensitive to the fluctuation of the collective threshold distribution and displays critical phenomena. By comparison, the equilibrium state with considering utility and psychological threshold looks positively robust. Furthermore, we observe that both cases demonstrate group bi-polarization pattern with the increase of standard deviation of the threshold.

Zhenpeng Li, Xijin Tang

Recognition of Online Opinion Leaders Based on Social Network Analysis

Opinion leaders on the internet play an important role in promoting the formation of online public opinion, which can influence the direction of public opinion. In this paper, we use social network analysis and content analysis to recognize the opinion leaders of online communities. First of all, we propose “Eight Degree” attribute indexes based on the characteristics of opinion leaders. Then, we construct an attribute matrix of the topic participants, and use the comprehensive evaluation to recognize opinion leaders. Finally, we divide the opinion leaders into the crucial active figure and the important position figure; we can also explore the potential opinion leader deeply by deleting the opinion leaders. The theoretical significance and practical value of this method have been demonstrated by a case study.

Ma Ning, Liu Yijun, Tian Ruya, Li Qianqian

Critical Infrastructure Management for Telecommunication Networks

Telecommunication network infrastructures such as cables, satellites, and cellular towers, play an important role in maintaining the stability of society worldwide. The protection of these critical infrastructures and their supporting structure become highly challenged to both public and private organizations. The understanding of interdependency of these infrastructures is the essential step to protect these infrastructures. This paper presents a critical infrastructure detection model to discover the interdependency based on the model from social network and new telecommunication pathways, while this study focuses on social theory into computational constructs. The policy and procedure of protecting critical infrastructures are discussed, and computational results from the proposed model are presented.

Haibo Wang, Bahram Alidaee, Wei Wang

Developing Self-Organizing Systems by Policy-Based Self-Organizing Multi-Agent Systems

Designing suitable interaction behaviors among agent to satisfy the system level requirements is a key issue when designers develop self-organizing systems. Such system is typically designed in an iterative way. However, this method is insufficient once the system requirements become unpredicted at the design-time. In this paper, we propose a policy-based approach to deal with this issue. In our approach, policy is the high-level abstraction to specify self-organizing mechanism as well as the mediator to adjust the local interaction behavior of agents, which enables developer to encapsulate self-organizing implementation mechanism and adapt to the changes of self-organizing requirements. We also study the corresponding implementation approach based on agent technology, including policy management and implementation, the corresponding agent architecture. Based on these implementation technologies, a running environment is provided to support the development of such system. Finally, a case is studied by the running environment to illustrate the effectiveness of our approach.

Yi Guo, Xinjun Mao, Cuiyun Hu, Junwen Yin, Xinzhong Zhu

On Prioritized 2-tuple Ordered Weighted Averaging Operators

This paper deals with linguistic aggregation problems where there exists a prioritization relationship over attributes. We propose a prioritized 2-tuple ordered weighted averaging (PTOWA) operator and study its properties. We then use this operator and a TOWA operator to aggregate satisfactions of attributes for alternatives.

Cuiping Wei, Xijin Tang, Xiaojie Wang

Special Session on Social Knowledge Discovery and Management

A Novel Collaboration Partner Model Based on the Personal Relationships of SNS

In this paper, we describe a novel model for locating appropriate ‘helpers’ for users based on the Chain of Friends (CoF) personal relationship in a SNS system, in order to locate appropriate ‘helpers’ for different users. This model is called SESNMM (Search Engine for Social Networked Mobile Model) and allows individual users, located in remote locations, to participate in a collaborative online community, via our SESNMM-based system. Such typical helpers are willing to help other users solve their tasks/problems and it is intended that both the users and helpers gain knowledge from these interactive online sessions. We have applied this model for inviting PC members of an international conference – namely LTLE 2012. The results showed that our model is very effective for discovering collaboration partners, locating useful helpers, finding users with similar interests in order to create communities for providing future and longer-term helping and teaching exchange.

Chengjiu Yin, Jane Yin-Kim Yau, Yoshiyuki Tabata

An Innovative Way for Mining Clinical and Administrative Healthcare Data

A novel method of “predicting” sitter case attribute value is presented in this paper. The method allows users to choose two attributes, seed and target attribute, and to predict the target attribute value of the forthcoming sitter case. The method first retrieves string sequences of the seed attribute according to filters the users set. Then, it finds the words in the sequences and calculates the term frequencies of the words. With the term frequencies, the proposed method uses vector space model to measure the similarity between the testing sequences and the benchmark sequence. At the end, the testing sequence which has highest Cosine similarity value is chosen and the filtering value the method uses to generate the testing sequence is the predicted result. These predicted results allow hospitals to adjust their strategies on resource assignments to better handle patient needs.

Siu Hung Keith Lo, Maiga Chang

An Adaptive Recommendation System for Museum Navigation

People like to attend exhibition activities, but hard to enter into the information effectively. We build new system with wireless internet and mobile device to guide visitor into the core information initiatively and effectively. The mobile guide system could classify visitor base on exhibition information and personal information that provide more suitable for users. Our system combined with semantic web technology to connect items data which users’ markup the type or property information in our system to created human portfolio. Our system is in compliance with human portfolio and metadata method to provide user information automatically and appropriately.

Jason C. Hung, Chun-Hong Huang, Victoria Hsu

Adaptive SVM-Based Classification Systems Based on the Improved Endocrine-Based PSO Algorithm

In this study, we proposed an wrapped feature selection and SVM’s kernel parameters optimization scheme using Improved Artificial Endocrine System to get an optimal support vector machines classification system. By taking the advantage of the mechanisms of hormone action in Artificial Endocrine System, we can avoid to obtain local optimums and oscillations. We used the UCI database to evaluate the performance of the proposed scheme with the previous methods. The experiment results indicated that the proposed scheme can avoid local optimum and also reduce feature numbers significantly with a good-enough accuracy in high-complexity datasets. Moreover, by decreasing the number of unnecessary features, we can even improve the accuracy of classification.

Kuan-Cheng Lin, Sheng-Hwa Hsu, Jason C. Hung

A Probability Model for Recognition of Dynamic Gesture Based on a Finger-Worn Device

Gesture recognition based on body-worn devices can be used for healthy improvement and life support. Many methods have been proposed for gesture recognition. However, most of them are concerned about recognition accuracy only. In some practical applications, real-time performance of a recognition method on a wearable device is also a key problem. In the paper, we propose a probability model for accurate and real-time recognition of dynamic gestures. The model learns from HMM but difference in the sense that our model builds probability matrices of feature distribution for each observation points instead of each gesture, which reduces the number of probability matrix to improve processing efficiency. A gesture can be recognized by the way of look-up table to search maximum similarity to pre-stored gestures in the matrices. To verify the model, eight kinds of one-stroke finger gestures are taken as the target of recognition. Result shows reasonable recognition accuracy and computational complexity.

Yinghui Zhou, Zixue Cheng, Lei Jing, Junbo Wang

Design of a Situation-Aware System for Abnormal Activity Detection of Elderly People

Internet of Things (IoT) is becoming one of hottest research topics. Elderly care is one of important applications in IoT, to grasp the situations around the elder people and then corresponding information can be sent to the care-givers to support the elder people. Abnormal activity detection is a particularly important task in the field, since the services should be immediately provided in such cases. Otherwise the elder people may be in danger. The existing approaches to this problem use some basic living patterns of the elder people, e.g. mobility per day, to detect abnormal activities. However, the detail abnormal activities in various specific situations cannot be detected, e.g., whether there is some abnormal activity when the elder people go to toilet, sleeps or eats something. To solve the above problem, in the paper, we propose a situation-aware abnormality detection system based on SVDD for the elder people. An experiment has been performed focusing on feasibility of the method and accuracy of the system to detect situations and abnormities from real sensors.

Junbo Wang, Zixue Cheng, Mengqiao Zhang, Yinghui Zhou, Lei Jing

Special Session on Human-Computer Interaction and Knowledge Discovery from Big Data

Revealing Cultural Influences in Human Computer Interaction by Analyzing Big Data in Interactions

Understanding human information needs worldwide requires the analysis of much data and adequate statistical analysis methods. Big interaction data from empirical studies regarding cultural human computer interaction (HCI) was analyzed using statistical methods to develop a model of culturally influenced HCI. There are significant differences in HCI style depending on the cultural imprint of the user. Having knowledge about the relationship between culture and HCI using this model, the local human information needs can be predicted for a worldwide scope.

Rüdiger Heimgärtner, Harald Kindermann

Predicting Student Exam’s Scores by Analyzing Social Network Data

In this paper, we propose a novel method for the prediction of a person’s success in an academic course. By extracting log data from the course’s website and applying network analysis methods, we were able to model and visualize the social interactions among the students in a course. For our analysis, we extracted a variety of features by using both graph theory and social networks analysis. Finally, we successfully used several regression and machine learning techniques to predict the success of student in a course. An interesting fact uncovered by this research is that the proposed model has a shown a high correlation between the grade of a student and that of his “best” friend.

Michael Fire, Gilad Katz, Yuval Elovici, Bracha Shapira, Lior Rokach

SPTrack: Visual Analysis of Information Flows within SELinux Policies and Attack Logs

Analyzing and administrating system security policies is difficult as policies become larger and more complex every day. The paper present work toward analyzing security policies and sessions in terms of security properties. Our intuition was that combining both visualization tools that could benefit from the expert’s eyes, and software analysis abilities, should lead to a new interesting way to study and manage security policies as well as users’ sessions. Rather than trying to mine large and complex policies to find possible flaws within, work may concentrate on which potential flaws are really exploited by attackers.

Actually, the paper presents some methods and tools to visualize and manipulate large SELinux policies, with algorithms allowing to search for paths, such as information flows within policies.

The paper also introduces a complementary original approach to analyze and visualize real attack logs as session graphs or information flow graphs, or even aggregated multiple-sessions graphs.

Our wishes is that in the future, when those tools will be mature enough, security administrator can then confront the statical security view given by the security policy analysis and the dynamical and real-world view given by the parts of attacks that most often occurred.

Patrice Clemente, Bangaly Kaba, Jonathan Rouzaud-Cornabas, Marc Alexandre, Guillaume Aujay

Using Mixed Node Publication Network Graphs for Analyzing Success in Interdisciplinary Teams

Large-scale research problems (e.g. health and aging, eonomics and production in high-wage countries) are typically complex, needing competencies and research input of different disciplines [1]. Hence, cooperative working in mixed teams is a common research procedure to meet multi-faceted research problems. Though, interdisciplinarity is – socially and scientifically – a challenge, not only in steering cooperation quality, but also in evaluating the interdisciplinary performance. In this paper we demonstrate how using mixed-node publication network graphs can be used in order to get insights into social structures of research groups. Explicating the published element of cooperation in a network graph reveals more than simple co-authorship graphs. The validity of the approach was tested on the 3-year publication outcome of an interdisciplinary research group. The approach was highly useful not only in demonstrating network properties like propinquity and homophily, but also in proposing a performance metric of interdisciplinarity. Furthermore we suggest applying the approach to a large research cluster as a method of self-management and enriching the graph with sociometric data to improve intelligibility of the graph.

André Calero Valdez, Anne Kathrin Schaar, Martina Ziefle, Andreas Holzinger, Sabina Jeschke, Christian Brecher

On Text Preprocessing for Opinion Mining Outside of Laboratory Environments

Opinion mining deals with scientific methods in order to find, extract and systematically analyze subjective information. When performing opinion mining to analyze content on the Web, challenges arise that usually do not occur in laboratory environments where prepared and preprocessed texts are used. This paper discusses preprocessing approaches that help coping with the emerging problems of sentiment analysis in real world situations. After outlining the identified shortcomings and presenting a general process model for opinion mining, promising solutions for language identification, content extraction and dealing with Internet slang are discussed.

Gerald Petz, Michał Karpowicz, Harald Fürschuß, Andreas Auinger, Stephan M. Winkler, Susanne Schaller, Andreas Holzinger

Human Involvement in Designing an Information Quality Assessment Technique – Demonstrated in a Healthcare Setting –

Information quality (IQ) has gained increasing importance in the last decade, yet results of assessing and improving IQ in practice are still rare. In this paper we employ a human-centered design approach and illustrate how human involvement in designing an IQ assessment technique can improve the resulting IQ assessment technique. We demonstrate the engagement with practitioners, users, and researchers during the design stages of the assessment technique. Using an emergency medical care (EMS) case our human-centered design approach is scoped to a healthcare setting. The design approach resulted in an improved assessment technique that assists increasing the quality of information exchanges. Our results showed the importance and impacts of human involvement during a design process.

Shuyan Xie, Markus Helfert, Lukasz Ostrowski

On Applying Approximate Entropy to ECG Signals for Knowledge Discovery on the Example of Big Sensor Data

Information entropy as a universal and fascinating statistical concept is helpful for numerous problems in the computational sciences. Approximate entropy (ApEn), introduced by Pincus (1991), can classify complex data in diverse settings. The capability to measure complexity from a relatively small amount of data holds promise for applications of ApEn in a variety of contexts. In this work we apply ApEn to ECG data. The data was acquired through an experiment to evaluate human concentration from 26 individuals. The challenge is to gain knowledge with only small ApEn windows while avoiding modeling artifacts. Our central hypothesis is that for intra subject information (e.g. tendencies, fluctuations) the ApEn window size can be significantly smaller than for inter subject classification. For that purpose we propose the term


to complement the statistical validity of a distribution, and show how truthfulness is able to establish trust in their local properties.

Andreas Holzinger, Christof Stocker, Manuel Bruschi, Andreas Auinger, Hugo Silva, Hugo Gamboa, Ana Fred

Towards a Framework Based on Single Trial Connectivity for Enhancing Knowledge Discovery in BCI

We developed a framework for systematic evaluation of BCI systems. This framework is intended to compare features extracted from a variety of spectral measures related to functional connectivity, effective connectivity, or instantaneous power. Different measures are treated in a consistent manner, allowing fair comparison within a repeated measures design. We applied the framework to BCI data from 14 subjects recorded on two days each, and demonstrated the framework’s feasibility by confirming results from the literature. Furthermore, we could show that electrode selection becomes more focal in the second BCI session, but classification accuracy stays unchanged.

Martin Billinger, Clemens Brunner, Reinhold Scherer, Andreas Holzinger, Gernot R. Müller-Putz


Weitere Informationen

Premium Partner