Mobile users can open program in new tab for better viewing.

Day 1 19/08/2019
Room #1

Day 2 20/08/2019
Room #1

## Keynote Speech I: Tasos Dagiuklas 08:20 - 09:10

Title: Challenges and Opportunities in 5G and Beyond

## Keynote Speech II: Jun Xu 09:10 - 10:00

Title: The Past and Future Impacts of Cloud Computing

## Panel Discussion: RESCUE-2019: Civil Protection Volunteers Training (CiProVoT) 10:30 - 11:30

Tasos Dagiuklas, Muddesar Iqbal, Tom Hales, Richard Abbot, Ian Greatbatch, Raheela Saad

Muddesar Iqbal

Shancang Li

## Tutorial III: Protection of Critical National Infrastructures 14:40 - 15:30

Leandros Maglaras
Day 3 21/08/2019
Room #1

## The Cruise Gala Dinner 17:15 - 20:15

Aboard the CityCruises boat Westminster
Room #2

## SESSION 1A - Cloud, IoT & Edge Computing (Nelson Haden Lecture Theatre, Faraday Wing Building) 07:00 - 08:30 ↓↑

Session Chair: Tong Liu
07:00 - 08:30
Accuracy-Guaranteed Event Detection via Collaborative Mobile Crowdsensing with Unreliable Users

Recently, mobile crowdsensing has become a promising paradigm to collect rich spatial sensing data, by taking advantage of widely distributed sensing devices like smartphones. Based on sensing data, event detection can be conducted in urban areas, to monitor abnormal incidents like traffic jam. However, how to guarantee the detection accuracy is still an open issue, especially when unreliable users who may report wrong observations are considered. In this work, we focus on the problem of user recruitment in collaborative mobile crowdsensing, aiming to optimize the fine-grained detection accuracy in a large urban area. Unfortunately, the problem is proved to be NP-hard, which means there is no polynomial-time algorithm to achieve the optimal solution unless P $=$ NP. To meet the challenge, we first employ a probabilistic model to characterize the unreliability of users, and measure the uncertainty of inferring event occurrences given collected observations by Shannon entropy. Then, by leveraging the properties of adaptive monotonicity and adaptive submodularity, we propose an adaptive greedy algorithm for user recruitment, which is theoretically proved to achieve a constant approximation ratio guarantee. Extensive simulations are conducted, which show our proposed algorithm outperforms baselines under different settings.
Authors: Tong Liu (Shanghai University), Wenbin Wu (Shanghai University), Yanmin Zhu (Shanghai Jiao Tong University), Weiqin Tong (Shanghai University),
Hide Authors & Abstract

Show Authors & Abstract
07:00 - 08:30
A Dynamic Difficulty-Sensitive Worker Distribution Model for Crowdsourcing Quality Management

Crowdsourcing utilizes the intelligence of people to solve problems that are difficult for machines such as entity resolution, sentiment analysis and image recognition. In crowdsourcing systems, requesters publish tasks that are answered by workers. However, the responses collected from the crowd are ambiguous as the workers on internet with unknown and very diverse abilities, skills, interests and knowledge background. In order to ensure the quality of crowdsourcing results, it is important to characterize worker quality accurately. Many previous works model the worker quality by a fixed value (such as probability value or confusion matrix). But even when workers complete the same type of tasks, the quality is affected by some factors (task difficulty) to varying degrees. Here we propose a dynamic difficulty-sensitive worker quality distribution model. In our model, the worker's ability is affected by task difficulty and fits a functional distribution. This model reflects the relationship between worker reliability and task difficulty. In addition, we utilize Expectation-Maximization approach (EM) to obtain maximum likelihood estimates of the parameters of worker quality distribution model and the true answers to the tasks. We conduct extensive experiments with synthetic data and real-world data. The experimental results show that our method significantly outperforms other state-of-the-art approaches.
Authors: Miao Zheng, Lizhen Cui (School of Software,Shandong University,Jinan,China), Wei He (School of Software,Shandong University,Jinan,China), Wei Guo (School of Software,Shandong University,Jinan,China), Xudong Lu (School of Software,Shandong University,Jinan,China),
Hide Authors & Abstract

Show Authors & Abstract
07:00 - 08:30
Priority-based Optimization of I/O Isolation for Hybrid Deployed Services

With the increasing of software complexity and user demands, collaborative service is becoming more and more popular. Each service focuses on its own specialty, their cooperation can support complicated task with high efficiency. To improve the resources utilization, virtualization technology like container is used and it enables multiple services running in the same physical machine. However, since the host physical machine is shared by several services, the resource competition is inevitable. Isolation is an effective solution, but the weak isolation mechanisms of container cannot handle such complicated scenarios. In the worst situation, the performance of services cannot meet the requirements and the system may crash. In order to solve this problem, we propose a priority-based optimization mechanism for I/O isolation after analyzing the characteristics of typical service workloads. Based on the real-time performance data, priority is automatically assigned to each service and corresponding optimization methods are applied. We evaluate the optimization effects of the priority-based mechanism in both static and dynamic workload cases, besides, the influence of different priority order is also analyzed. The experimental results show that our approach can indeed improve the system performance and guarantee the requirements of all the running services are satisfied.
Authors: Jiancheng Zhang (Hangzhou Dianzi University), Youhuizi Li (Hangzhou dianzi University), Li Zhou (Hangzhou Dianzi University), Zujie Ren (Zhejiang Lab), Jian Wan (Zhejiang LabZhejiang University of Science and Technology), Yuan Wang (NetEase Hangzhou),
Hide Authors & Abstract

Show Authors & Abstract
07:00 - 08:30
Optimal Device Management Service Selection in Internet-of-Things

In Internet-of-Things(IoT), IoT device management is a challenge for device owners considering the huge amount of devices and their heterogeneous quality of service(QoS) requirements. Recently, IoT device management service(MS) providers are arising to serve device owners. Device owners can now easily manage their devices by using IoT device MSs. It is critical to select suitable MSs from numerous candidates for devices. An optimal service selection must maximize the number of MS managed devices and minimize the total cost while ensuring the QoS requirements of IoT system. To optimize the IoT Device Management Service Selection problem, we propose IDMSS, a Lexicographic Goal Programming(LGP) based approach. However, due to the high computational complexity of the IoT Device Management Service Selection problem, an alternative heuristic-based approach called GA4MSS is proposed. Two series of experiments have been conducted and the experimental results show the performance of our approaches.
Authors: Weiling Li (Chongqing University), Yunni Xia (Chongqing University), Wanbo Zheng (Kunming University of Science and Technology), Peng Chen (Xihua University), Jia Lee (Chongqing University), Yawen Li (Chongqing University),
Hide Authors & Abstract

Show Authors & Abstract
07:00 - 08:30
A Dynamic Planning Framework for QoS-based Mobile Service Composition under Cloud-Edge Hybrid Environments

In cloud-edge hybrid environments, when QoS constraints of the SOA-based mobile service composition change, a dynamic reconfiguration needs to be performed. Different from traditional cloud service, the cloud-edge hybrid environment has the characteristics of limited resource storage, limited energy at the edge and uncertain users who move frequently. Dynamic reconfiguration in this mode is challenging. QoS is an important indicator of service evaluation. Most studies focus on only the static QoS attributes of the service. However, the QoS of a service is not statically constant; it changes dynamically over time. Therefore, to avoid the immediate failure of the service and ensure the stability of the mobile service composition after dynamic reconfiguration, an LSTM neural network is used to predict the future value of QoS for candidate service, which is used as an evaluation indicator for the service during dynamic reconfiguration. Then, attributes such as energy consumption, traffic and moving track are considered. A cost-reward mechanism is constructed to calculate the cost and reward of the service when it is invoked. The reasonable restriction conditions are added for controlling dynamic reconfiguration. Finally, the dynamic reconfiguration problem-solving process and framework for mobile service composition based on QoS in a cloud-edge hybrid environment is introduced, guiding the mobile service composition dynamic reconfiguration task in cloud-edge hybrid environments.
Authors: honghao gao (School of Computer Engineering and Science, Shanghai University, Shanghai, China), wanqiu huang (School of Computer Engineering and Science, Shanghai University, Shanghai, China), qiming zou (Computing Center, Shanghai University, Shanghai, China), xiaoxian yang (School of Computer and Information Engineering, Shanghai Polytechnic University),
Hide Authors & Abstract

Show Authors & Abstract
07:00 - 08:30
A Security Framework to Protect Edge Supported Software Defined Internet of Things Infrastructure

Managing the huge IoT infrastructure poses a vital challenge to the network community. Software Defined Networking (SDN), due to its characteristics of centralized network management has been considered as an optimal choice to manage IoT. Edge computing brings cloud recourses near the IoT to localize the cloud demands. Hence, SDN, IoT, and edge computing create a resourceful SDIoT-Edge architecture to efficiently orchestrate cloud services and utilize resource-limited IoT devices in a flexible way. Besides a wide adoption of IoT, the vulnerabilities present in this less secure infrastructure can be exploited by the adversaries to attack the OpenFlow channel using Distributed Denial of Service (DDoS) attacks. DDoS on OpenFlow channel have the ability to disrupt the whole network hence, providing security for the OpenFlow channel is a key challenge in SDIoT-Edge. We propose a security framework called SDIoT-Edge Security (SIESec) against the security vulnerabilities present in this architecture. SIESec prototype employs machine learning-based classification strategy, blacklist integration, and contextual network flow filtering to efficiently defend against the DDoS attacks. We perform extensive simulations using Floodlight controller and Mininet network emulator. Our results proclaim that SIESec provides extensive security against OpenFlow channel DDoS attacks and pose a very less overhead on the network.
Authors: Wajid Rafique (Nanjing University, Nanjing, P. R. China), Wanchun Dou (Nanjing University, Nanjing, P. R. China), Nadeem Sarwar (Bahria University, Lahore, Pakistan), Maqbool Khan (Nanjing University, P. R. China),
Hide Authors & Abstract

Show Authors & Abstract

## SESSION 2A - Collaborative IoT and Services Applications (Nelson Haden Lecture Theatre, Faraday Wing Building) 11:00 - 13:00 ↓↑

Session Chair: Yueshen Xu
11:00 - 13:00
An Integrated and Intelligent Dental Healthcare System with Mobile Services

Medical informatization takes a key role in medical and healthcare industries, which is a necessary way of improving service quality and treatment experience in a hospital. In this paper, we design and implement an integrated and intelligent healthcare system with mobile services, specific to dental healthcare. The developed system contains four components, including WeChat official account, intelligent question and answering (Q&A) system, mobile follow-up care mini program and AI speech assistant. The developed WeChat official account and intelligent Q&A system provide a large amount of professional knowledge on dental health, which can help narrow the knowledge gap between doctors and patients. The two components facilitate patients to follow up our applications without downloading any extra softwares, as the developed functions (e.g., voice service) are provided through mobile services. These two components also facilitate follow-up management, decreasing manpower and resource usage in hospital. Our system is developed to serve patients as well as dentists, and provides a group of interactive healthcare services in a low cost. In this paper, we elaborate the whole system architecture and implementation detail of each component. We also report the performance test result and training process of the classifier.
Authors: Yueshen Xu (Xidian University), Yi Lu (Stomatological Hospital, Xi'an Jiaotong University), Rui Li (Xidian University), Lin Niu (Stomatological Hospital, Xi'an Jiaotong University), Wenzhi Du (Stomatological Hospital, Xi'an Jiaotong University), Ni An (Stomatological Hospital, Xi'an Jiaotong University), Yaning Liu (Xidian University), Xinyi Liu (Xidian University), Yan Jiang (Xidian University), Zhenhua Li (Xidian University), Jin Guo (Xidian University), Xiangdong Wang (Shaanxi Lishengruihui Technology Co., Ltd),
Hide Authors & Abstract

Show Authors & Abstract
11:00 - 13:00
An Edge Computing-based Framework for Marine Fishery Vessels Monitoring Systems

Vessel Monitoring Systems (VMS) have been adopted by many countries which provide information on the spatial and temporal distribution of fishing activity. Real-time communication and interaction between fishing vessels and shore-based systems is a weakness of traditional vessel monitoring systems. This paper proposes a novel framework of edge computing-based VMS (EC-VMS). The framework of EC-VMS mainly consists of four layers. An edge computing terminal is used on each vessel, and the BeiDou navigation satellite system (BDS) is adopted for communication. Meanwhile, edge computing servers interact with corresponding management vessels and the cloud. In order to decrease the communication cost, a data transmission policy called Adaptable Trajectory Transmission Model (ATTM) is presented in this paper. The experimental results illustrate the efficiency of the proposed EC-VMS, with the average communication time greatly decreased in a typical scenario. Moreover, EC-VMS improves the real-time performance of the system.
Authors: Fengwei Zhu (School of Computer Science and Technology, Hangzhou Dianzi University), Yongjian Ren (School of Computer Science and Technology, Hangzhou Dianzi University), Jie Huang (School of Computer Science and Technology, Hangzhou Dianzi University), Jian Wan (School of Computer Science and Technology, Hangzhou Dianzi University), Hong Zhang (Institute of Science and Technology Information Research of Zhejiang Province),
Hide Authors & Abstract

Show Authors & Abstract
11:00 - 13:00
A Mobile and Web-based Approach for Targeted and Proactive Participatory Sensing

Participatory sensing applications have gained popularity due to the in-creased use of mobile phones with embedded sensors. One of the main is-sues in participatory sensing applications is the uneven coverage of areas, i.e., some areas might be covered by multiple participants while there is no data for other areas. In this paper, we design mobile and web-based infra-structure to enable domain scientists to effectively acquire crowd-sensed data from specific areas of interest (AOIs) to support the goal of even cover-age for data collection. Scientists can mark the AOIs on a web-portal, then volunteers will be proactively informed about the participatory sensing opportunities near their current location. We present a caching algorithm to increase the performance of our proposed system and studied the perfor-mance of the caching algorithm for different real-world scenarios on differ-ent mobile phones. We observed that prefetching data improves the perfor-mance to some extent; however, it starts to degrade after a certain point de-pending upon the number of nearby AOIs.
Authors: Navid Hashemi Tonekaboni (University of Georgia), Lakshmish Ramaswamy (University of Georgia), Sakshi Sachdev (University of Georgia),
Hide Authors & Abstract

Show Authors & Abstract
11:00 - 13:00
Forecasting Long-term Call Traffic based on Seasonal Dependencies

How to use future call traffic for scheduling different staffs to work in a month or a week is an important task for call center. In this problem setting, the call traffic should be predicted in a long-term way where the forecasting results for different periods are required. However, it is very challenging to solve this problem due to the randomness nature of the call traffic and the multiple forecasting in long term. Current methods cannot solve this problem since they either merely focus on short-term forecasting for the next hour or next day, or ignore call-holding time for call traffic prediction. In this paper, we propose an effective method for predicting long-term call traffic with multiple forecasting results for different future periods, e.g., every 15 minutes, and take both call arrival rate and call-holding time into consideration through the Erlang. In our method, the seasonal dependencies are summarized by performing data analysis, then different features based on these dependencies are extracted for training the prediction model.In order to forecast call traffic of multiple time buckets, we propose two strategies based on different features. The evaluation results show that the features, the prediction models and the strategies are feasible.
Authors: Cao Longchun (Zhejiang University of Technology), Ma Kui (Zhejiang University of Technology), Cao Bin (Zhejiang University of Technology), Fan Jing (Zhejiang University of Technology),
Hide Authors & Abstract

Show Authors & Abstract
11:00 - 13:00
Attention-based Bilinear Joint Learning Framework for Entity Linking

Entity Linking (EL) is a task that links entity mentions in the text to corresponding entities in a knowledge base. The key to building a high-quality EL system involves accurate representations of word and entity. In this paper, we propose an attention-based bilinear joint learning framework for entity linking. First, a novel encoding method is employed for coding EL. This method jointly learns words and entities using an attention mechanism. Next, for ranking features, a weighted summation model is introduced to model the textual context and coherence. Then, we employ a pairwise boosting regression tree (PBRT) to rank candidate entities. As input, PBRT takes both features constructed with a weighted summation model and conventional EL features. Finally, through the experiment, we demonstrate that the proposed model learns embedding efficiently and improves the EL performance compared with other state-of-the-art methods. Our approach achieves superior result on two standard EL datasets: CoNLL and TAC 2010
Authors: Min Cao (Shanghai University), Penglong Wang (Shanghai University), Honghao Gao (Shanghai University), Jiangang Shi (Shanghai Shang Da Hai Run Information System Co), Yuan Tao (Shanghai University), Weilin Zhang (Shanghai University),
Hide Authors & Abstract

Show Authors & Abstract
11:00 - 13:00
A Collaborative Anomaly Detection Approach of Marine Vessel Trajectory

Trajectory anomaly detection plays a very important role in safety management of navigation. Most trajectory anomaly detection methods mainly detect the spatial information of vessel's trajectory. These methods neglect vessel's dynamic behavior characteristics such as course, speed and acceleration. In this paper, a Vessel Trajectory Multi-factor Collaborative Anomaly Detection (VT-MCAD) approach is proposed to realize anomaly detection of vessels at sea by studying trajectory characteristics. Firstly, the trajectory behavior of historical vessels is identified, and the trajectory characteristics such as heading, speed and acceleration are extracted for different trajectory behaviors. Then, the current trajectory behavior is identified when the trajectory anomaly is detected. On the basis of TRAOD method, the corresponding trajectory feature model components such as heading, speed and acceleration are used to detect the trajectory anomaly, and the trend points of the trajectory anomaly are obtained. Finally, the outlier trend score of each component is combined to calculate the final outlier trend score. VT-MCAD can change the weight of components according to the detection effectiveness of different components and avoid excessive dependence on one component, so it has better robustness and reliability. The experimental results on real-world vessel data show that VT-MCAD can effectively capture abnormal trajectories.
Authors: Zejun Huang (Hangzhou Dianzi University), Jian Wan (Hangzhou Dianzi University), Jie Huang (Hangzhou Dianzi University), Gangyong Jia (Hangzhou Dianzi University), Wei Zhang (Hangzhou Dianzi University),
Hide Authors & Abstract

Show Authors & Abstract
11:00 - 13:00
Classification of skin lesions based on data collaboration under imbalance dataset

Imbalance data is a common problem in machine learning task, which often impacts the accuracy of models. An effective way to solve this is to increase the number of minority class samples in the dataset. There are many methods to solve the problem of imbalance data in the field of machine learning. But these are all for low-dimensional data. For high-dimensional data, such as images, these methods are not well applicable. In this paper, an image generation method based on generative adversarial network is introduced to carry out pattern learning for samples of minority class in the dataset, so as to realize the expansion of data for minority class. And finally the classification network for skin lesions is trained by data collaboration which consist of real images and generated images. The experimental results indicate that the addition of generated images further improves the accuracy of the network while alleviating the imbalance problem to some extent.
Authors: Weijia Ji (School of Information Science and Engineer, East China University of Science and Technology, Shanghai, China), Lizhi Cai (Laboratory of Computer Software Testing & Evaluating, Shanghai Development Center of Com-puter Software Technology, Shanghai, China), Mingang Chen (Laboratory of Computer Software Testing & Evaluating, Shanghai Development Center of Com-puter Software Technology, Shanghai, China), Naiqi Wang (School of Information Science and Engineer, East China University of Science and Technology, Shanghai, China),
Hide Authors & Abstract

Show Authors & Abstract
11:00 - 13:00
Wi-Fi Imaging Based Segmentation and Recognition of Continuous Activity

Automatic segmentation and action recognition have been a long-standing problem in sensorless sensing. In this paper, we propose CHAR, a continuous human activity recognition system to solve these problems in a different way. We’ve noticed that these challenges have been solved in image processing field, so CHAR could effectively perform action segmentation and recognition after WiFi imaging. The key idea behind Wi-Fi imaging is that different body part reflects transmitted signal, the receiver receives the combination of them, and then we separate the received signals from different directions and get the signal intensity in each direction to get the heat map showing the shape of the object. The imaging sequence contains multiple pictures records a continuous action at different time, and we can easily separate and recognize the action based on $IC^{2}$(image classification), a classification framework we proposed. We implement CHAR using commodity WiFi devices to evaluate its performance under different environment. The results show that the imaging result is better than prior works, facilitating CHAR to achieving an average recognition accuracy, i.e., > 95%.
Authors: Yang Zi (Xi'an Jiaotong University), Wei Xi (Xi'an Jiaotong University), Li Zhu (Xi'an Jiaotong University), Fan Yu (Xi'an Jiaotong University), Kun Zhao (Xi'an Jiaotong University), Zhi Wang (Xi'an Jiaotong University),
Hide Authors & Abstract

Show Authors & Abstract

## SESSION 3A - Security and Trustworthy (Nelson Haden Lecture Theatre, Faraday Wing Building) 13:30 - 14:30 ↓↑

Session Chair: Youhuizi Li
13:30 - 14:30
Application and Implementation of Multivariate Public Key Cryptosystem in Blockchain

Blockchain is one of the most revolutionary and innovative technologies in recent years. The traditional asymmetric encryption algorithms guarantee the security of data on blockchain. However, with the rapid development of quantum computing technologies, as long as large-scale quantum computers appear, these kind of encryption systems can be deciphered by shor algorithm in polynomial time. Therefore, blockchain technologies are going to face potential security threats. To solve this problem, the best solution at present is to replace the asymmetric encryption algorithms in the blockchain with post-quantum cryptosystems. In this paper, we apply the Rainbow algorithm with high signature efficiency to the existing Ethereum platform, and test the feasibility of the scheme by building a private chain. In addition, we compare the signature efficiency of Rainbow algorithm with ECDSA, which is expected to provide direction and inspiration for future research on blockchain resistance to quantum computing.
Authors: Ruping Shen (Chongqing University), Hong Xiang (Chongqing University), Xin Zhang (Chongqing University), Bin Cai (Chongqing University), Tao Xiang (Chongqing University),
Hide Authors & Abstract

Show Authors & Abstract
13:30 - 14:30
Secure Sharing Model Based On Block Chain in Medical Cloud

The cloud storage and sharing system are widely used in medical systems. The unique characteristics of cloud storage enable healthy data being efficiently delivered and retrieved. Nevertheless, traditional medical cloud system suffers two flaws. For one thing, centralized cloud servers are vulnerable to malicious attack and single point of failure. For another, these systems cannot offer a powerful capability to protect medical health data. Blockchain technology is considered to be one of vital technologies of Bitcoin. In this paper, we propose a security model which combines the cloud storage technology with blockchain technology. Our model adopts Delegate Proof of Stake (DPOS) consensus mechanism to ensure that all nodes have unified state in the network. Additionally, the CP-ABE scheme is introduced into the Proxy Re-encryption to store and share medical data which supports the keywords searching. Moreover, we rank medical institutions that different ranks have different duties. In our secure sharing models, there are no central nodes and it is a distributed environment. It not only can reduce the access overhead of the blockchain but also better resist the collusion attack. Furthermore, our security analysis indicates that the proposed scheme achieves provable security under the q-DBDHE assumption in the random oracle model. Then, the comparisons show that our model is more efficient and practical than previous ones.
Authors: Tao Feng (Lanzhou university of technology), ying jiao (Lanzhou university of technology),
Hide Authors & Abstract

Show Authors & Abstract
13:30 - 14:30
A Smart Topology Construction Method for Anti-tracking Network based on the Neural Network

Anti-tracking network is the effective method to protect the network users' privacy confronted with the increasingly rampant network monitoring and network tracing. But the architecture of the current anti-tracking network is easy to be attacked, traced and undermined. In this paper, We propose smart topology construction method (STon) to provide the self-management and self-optimization of topology for anti-tracking network. We firstly deploy the neural network on each node of the anti-tracking network. Each node can collect its local network state and calculate the network state parameters with the neural network to decide the link state with other nodes. At last, each node optimizes its local topology according to the link state. With the collaboration of all nodes in the network, the network can achieve the self-management and self-optimization of its own topology. The experimental results showes that STon has a better robustness, communication efficiency and anti-tracking performance than the current popular P2P structures.
Authors: Changbo Tian (School of Cyber Security, University of Chinese Academy of Sciences), YongZheng Zhang (Institute of Information Engineering, Chinese Academy of Sciences), Tao Yin (Institute of Information Engineering, Chinese Academy of Sciences), Yupeng Tuo (Institute of Information Engineering, Chinese Academy of Sciences), Ruihai Ge (Institute of Information Engineering, Chinese Academy of Sciences),
Hide Authors & Abstract

Show Authors & Abstract
13:30 - 14:30
Context-Aware Point-of-Interest Recommendation Algorithm with Interpretability

With the rapid development of mobile Internet, smart devices, and positioning technologies, location-based social networks (LBSNs) are growing rapidly. In LBSNs, point-of-interest (POI) recommendation is a crucial personalized location service that has become a research hotspot. To address extreme sparsity of user check-in data, a growing line of research exploits spatial-temporal information, social relationship, content information, popularity, and other factors to improve recommendation performance. However, the temporal and spatial transfers of user preferences are seldom mentioned in existing works, and interpretability, which is an important factor to enhance credibility of recommendation result, is overlooked. To cope with these issues, we propose a context-aware POI recommendation framework, which integrates users’ long-term static and time-varying preferences to improve recommendation performance and provide explanations. Experimental results over two real-world LBSN datasets demonstrate that the proposed solution has better performance than other advanced POI recommendation approaches.
Authors: Guoming Zhang (State Key Laboratory for Novel Software Technology, Nanjing University, Nanjing, China), Lianyong Qi (School of Information Science and Engineering, Qufu Normal University, Qufu, China), Xuyun Zhang (Department of Electrical Computer Engineering, The University of Auckland, Auckland, New Zealand), Xiaolong Xu (School of Computer and Software, Nanjing University of Information Science and Technology, Nanjing, China), Wanchun Dou (State Key Laboratory for Novel Software Technology, Nanjing University, Nanjing, China),
Hide Authors & Abstract

Show Authors & Abstract

## SESSION 4A - Algorithm, Network and Testbed (Nelson Haden Lecture Theatre, Faraday Wing Building) 14:30 - 15:30 ↓↑

Session Chair: Rafique Wajid
14:30 - 15:30
An Influence Maximization Algorithm Based on Real-time and De-superimposed Diffusibility

With the great development of the social network, exploring an influence maximi-zation algorithm with strong adaptability and superior performance will undoubt-edly produce great market value. Existing algorithms select a seed node with the greatest influence. This will inevitably have an influence in mutual coverage, which will have a more or less negative impact on the final results and reduce the performance of the algorithm. In this paper, Node Diffusibility is proposed, and it is updated it in real time and eliminated the deviation caused by its overlay. On the basis of traditional calculation of node influence, more attention was paid to the influence of a node's neighboring nodes rather than to the characteristics of the nodes themselves. The proposed algorithm was evaluated by experiments conducted on selected real data sets. Compared with the classical ranking-based algorithms, MaxDegree and PageRank, the proposed algorithm achieved better results in terms of efficiency and time complexity.
Authors: Yue Ren (Shanghai University), Xinyuan Zhang (Shanghai University), Liting Xia (Shanghai University), Yongze Lin (Shanghai University), Yue Zhao (Shanghai University), Weimin li (Shanghai University),
Hide Authors & Abstract

Show Authors & Abstract
14:30 - 15:30
Detecting Overlapping Communities of Nodes with Multiple Attributes from Heterogeneous Networks

Many methods have been proposed for detecting communities from heterogene-ous information networks with general topologies. However, most of these meth-ods can detect communities with homogeneous structures containing nodes with only a single attribute. Investigating methods for detecting communities contain-ing nodes with multiple attributes from heterogeneous information networks with general topologies has been understudied. Such communities are realistic in real-world social structures and exhibits many interesting properties. Towards this, we propose a system called DOMAIN that can detect overlapping communities of nodes with multiple attributes from heterogeneous information networks with general topologies. The framework of DOMAIN focuses on domains (i.e., attrib-utes) that describe human characteristics such as ethnicity, culture, religion, de-mographic, age, or the like. The ultimate objective of the framework is to detect the smallest sub-communities with the largest possible number of domains, to which an active user belongs. The smaller a sub-community is, the more specific and granular its interests are. The interests and characteristics of such a sub-community is the union of the interests and characteristics of the single domain communities, from which it is constructed. We evaluated DOMAIN by compar-ing it experimentally with three methods. Results revealed marked improvement
Authors: Kamal Taha (Khalifa University), Paul Yoo (University of London),
Hide Authors & Abstract

Show Authors & Abstract
14:30 - 15:30
Dynamical rating prediction with topic words of reviews: a hierarchical analysis approach

Social commerce is an important part of the social network which contains a large number of user behaviors and user relationships. Users generate reviews, social relations, user-product or product-product mapping information that can reflect an evolution of product characteristics and user preferences in using social commerce. It is a popular topic by using these information to conduct rating prediction in the field of intelligent recommendation. In this paper, optimizing the rating prediction based on topic analysis in two aspects.On the one hand, in the process of data preprocessing,constructing a dynamic hierarchical tree of topic words(DHTTW), which can not only capture the change of users' preferences for product property, but also reflect the impact of different product property on users' preferences at the same time. Based on DHTTW,designing the mapping rules from user reviews to DHTTW to generate user preference vectors. On the other hand, in the process of prediction,proposing a prediction method named combination of gradient boosting decision tree and multi-class linear regression(GBDT-MCLR), which further improves the accuracy of rating prediction.
Authors: Huibing Zhang (Guangxi Key Laboratory of Trusted Software,Guilin University of Electronic Technology, Guilin 541004, China.), Hao Zhong (Guangxi Key Laboratory of Trusted Software,Guilin University of Electronic Technology, Guilin 541004, China.), Qing Yang (Guangxi Key Laboratory of Automatic Measurement Technology and Instrument, Guilin University of Electronic Technology, Guilin 541004, China.), Fei Jia (Guangxi Key Laboratory of Trusted Software,Guilin University of Electronic Technology, Guilin 541004, China.), Ya Zhou (Guangxi Key Laboratory of Trusted Software,Guilin University of Electronic Technology, Guilin 541004, China.), Fang Pan (Teaching Affairs Office, Guangxi Normal University, Guilin 541004, China.),
Hide Authors & Abstract

Show Authors & Abstract
14:30 - 15:30
NTS: A Scalable Virtual Testbed Architecture with Dynamic Scheduling and Backpressure

Experimental platforms perform a key role in evaluating the proof-of- concept and innovations. Nowadays, researchers from academia and industries rely on expensive physical testbeds to evaluate their experiments,while there are very limited software testbeds in market, which usually not available or costly. In addition, the applications of existing traffic generators are restricted to their sin- gle function and performance in network area. It has come to a point that lack of validation and testing tools has tremendously jeopardized the innovation in this field. In this paper, we propose NTS, which is a scalable software-based virtual testbed architecture. The scheduling and management framework can dynamical- ly schedule resource of services. The scheduling algorithm adopts the concept of cost proportional fairness scheduling, which takes the evaluated traffic propor- tion and packet arrival rate into account. By leveraging container technology, the resources of services are restrictedly managed and fully isolated without tamper- ing the OS kernel’s scheduling mechanisms. Another advantage of the proposed testbed architecture is that the software can generate most kinds of backbone net- work traffic and can also be extended easily for customized protocol or traffic patterns. Our experiments show that the virtual testbed is generic scalable and cost-efficient, which is suitable and affordable for researchers in the field of net- work.
Authors: Youbing Zhong (Institute of Information Engineering, Chinese Academy of Sciences School of Cyber Security, University of Chinese Academy of Sciences), Zhou Zhou (Institute of Information Engineering), Da Li (Dept. of Electrical and Computer Engineering, University of Missouri-Columbia), Wenliang He (Institute of Information Engineering, Chinese Academy of Sciences), Chao Zheng (Institute of Information Engineering, Chinese Academy of Sciences), Qingyun Liu (Institute of Information Engineering, Chinese Academy of Sciences), Li Guo (Institute of Information Engineering, Chinese Academy of Sciences),
Hide Authors & Abstract

Show Authors & Abstract
Room #3

## SESSION 1B - Data Analysis & Recommendation (FW-114., Faraday Wing Building) 07:00 - 08:30 ↓↑

Session Chair: Weilong Ding
07:00 - 08:30
SMART: A Service-oriented Statistical Analysis Framework on Spatio-temporal Big Data

Spatio-temporal data is one of the most important assets in the context of smart cities. Spatio-temporal big data comes from a variety of sensor equipment, implies the state of urban operation, insight into the development trend. Due to the multi-dimensional characteristics and diverse analysis needs of large spatial-temporal data, data analysis based on large spatial-temporal data must take into account the large capacity, diversity and frequent changes of data, as well as the query, integration and visualization of data. This makes spatial and temporal data analysis more difficult. In order to simplify the analysis of spatio-temporal data, a service-oriented intelligent framework is proposed. Firstly, the concept of spatio-temporal data service is introduced into the framework, and several common spatio-temporal data service models are defined. Then, a spatio-temporal data service composition language based on BPEL is proposed to define analysis applications. We also developed a prototype tool to implement spatio-temporal data services on Hadoop. In order to prove the applicability of our method, we demonstrate the effectiveness of our work through a practical application-based study.
Authors: Jie Zhou (1.Data Engineering Institute, North China University of Technology Beijing 100144 2.Beijing Key Laboratory on Integration and Analysis of Large-scale Stream Data Beijing), Weilong Ding (1.Data Engineering Institute, North China University of Technology Beijing 100144 2.Beijing Key Laboratory on Integration and Analysis of Large-scale Stream Data Beijing), Zhuofeng Zhao (1.Data Engineering Institute, North China University of Technology Beijing 100144 2.Beijing Key Laboratory on Integration and Analysis of Large-scale Stream Data Beijing), Han Li (1.Data Engineering Institute, North China University of Technology Beijing 100144 2.Beijing Key Laboratory on Integration and Analysis of Large-scale Stream Data Beijing),
Hide Authors & Abstract

Show Authors & Abstract
07:00 - 08:30
Collaborative Contextual Combinatorial Cascading Thompson Sampling

We design and analyze collaborative contextual combinatorial cascading Thompson sampling (C4-TS). C4-TS is a Bayesian heuristic to address the cascading bandit problem in the collaborative environment. C4-TS utilizes posterior sampling strategy to balance the exploration-exploitation tradeoff and it also incorporates the collaborative effect to share information across similar users. Utilizing these two novel features, we prove that the regret upper bound for C4- TS is \tilde{O} (d(u + pmKT )), where d is the dimension of the feature space, u is the number of users, m is the number of clusters, K is the length of the recommended list and T is the time horizon. This regret upper bound matches the theoretical guarantee for UCB-like algorithm in the same settings. We also conduct a set of simulations comparing C4-TS with the state-of-the-art algorithms. The empirical results demonstrate the advantage of our algorithm over existing works.
Authors: Zhenyu Zhu (USTC), Liusheng Huang (USTC), Hongli Xu (USTC),
Hide Authors & Abstract

Show Authors & Abstract
07:00 - 08:30
Multi-Label Recommendation of Web Services with the Combination of Deep Neural Networks

With the increasing number of web services on the Internet, how to effectively classify and recommend service labels has become a research issue. It plays an important role in web service organization and management. However, the deficiency of current approaches is that they either recommend only a single label for a web service or a set of independent labels without order ranking that is still difficult for service providers to publish their web services. In this paper, together with label embedding techniques, we propose a novel approach for service multi-label recommendation using deep neural networks. Unlike traditional approaches, the predicted service labels of our approach not only satisfy the demands of service multi-label recommendation, but also provide the importance with an ordered label ranking. The experiments are conducted to validate the effectiveness on a large-scale dataset from ProgrammableWeb, involving 13,869 real-world Web services. The experimental results demonstrate that our approach for multi-label recommendation of web services outperforms the competing approaches in terms of multiple evaluation metrics.
Authors: Yanglan Gan (School of Computer Science and Technology, Donghua University), Yang Xiang (School of Computer Engineering and Science, Shanghai University; Shanghai Institute for Advanced Communication and Data Science, Shanghai University), Guobing Zou (School of Computer Engineering and Science, Shanghai University; Shanghai Institute for Advanced Communication and Data Science, Shanghai University), Huaikou Miao (School of Computer Engineering and Science, Shanghai University; Shanghai Key Laboratory of Computer Software Evaluating and Testing), Bofeng Zhang (School of Computer Engineering and Science, Shanghai University),
Hide Authors & Abstract

Show Authors & Abstract
07:00 - 08:30
An Approach for Item Recommendation Using Deep Neural Network combined with the Bayesian Personalized Ranking

This paper proposes a deep neural network model (SDAE-BPR) based on Stack Denoising Auto-Encoder and Bayesian Personalized Ranking for the problem of accurate product recommendation. First, we use the Stack Denoising Auto-Encoder (SDAE) as the input of the item’s rating data and obtain the hidden features after encoding. Second, the Bayesian personalized Ranking (BPR) method is used to learn the hidden feature vector of the corresponding item. This model can avoid the influence of the sparseness of the matrix. Therefore, this model achieves the effect of more accurate recommendations of items. Third, to reduce the cost of model training, a unique pre-training and fine-tuning strategy is proposed in the deep neural network. Finally, based on the Movielens 20M dataset, the results of the SDAE-BPR, a traditional item-based collaborative filtering model and a user-based collaborative filtering model are compared. It is shown that the SDAE-BPR has higher accuracy. This method improves the accuracy of parameter estimation and the efficiency of model training.
Authors: Zhongqin Bi (College of Computer Science and Technology, ShangHai University of Electric Power, Shanghai, China), Siming Zhou (College of Computer Science and Technology,ShangHai University of Electric Power), Xiaoxian Yang (School of Computer and Information Engineering, Shanghai Polytechnic University), Ping Zhou (College of Computer Science and Technology, ShangHai University of Electric Power, Shanghai, China), Jiale Wu (College of Computer Science and Technology, ShangHai University of Electric Power, Shanghai, China),
Hide Authors & Abstract

Show Authors & Abstract
07:00 - 08:30
Itinerary Recommendation for User Groups in Temporary Social Network

Temporary social networks have been used at hotels, concerts, theme parks, and sports arenas, where people with a common interest or activity form a social group for a short time. People conﬁned to such speciﬁc places or activities are allowed to join the temporary social networks using their main social network accounts (e.g., Foursquare, Facebook). Users registered for the same business/research conference may have common connections and thus may be willing to travel together in the conference city. Currently, renting cars to travel has become very common, since it helps improve users’ experience as well as save travel costs (e.g., renting and oil costs). Thus, we propose a group-wise itinerary planning framework to minimize the travel costs for each user in temporary social network. Experimental results obtained by using real data sets illustrate the eﬀectiveness of our proposed framework.
Authors: Jing Xia (Hangzhou Dianzi University), Yu Li (Hangzhou Dianzi Universtiy), Yuyu Yin (Hangzhou Dianzi University),
Hide Authors & Abstract

Show Authors & Abstract
07:30 - 08:30
Extracting Topics from Semi-structured Data for Enhancing Enterprise Knowledge Graphs

Unifying information across the organizational data silos that lack documentation, structure and automated semantic discovery has been of an intense interest in the recent years. Enterprise knowledge graph is a common tool of data integration and knowledge discovery and it has become a backbone to APIs that demand access to structured knowledge. A piece which was previously unnoticed in building enterprise knowledge graph, is adding an abstract layer of themes and concepts which is mapped to various documents stored as semi-structured files in databases. Augmenting enterprise knowledge graphs by concepts will help companies to find the trends in their data and get a holistic view over their entire data stores. Extracting topics from semi-structured data suffers from lack of corpus or description as its major challenge. In this research, we investigate the impact of self-supplementation of words and documents on probabilistic topic modeling upon semi-structured data. Another contribution of this paper is finding the best tuning of probabilistic topic modeling that fits semi-structured data. We consider 2 inferencing techniques and demonstrate the results on real data pools from Open City data and Kaggle data containing 7.5 GB and 1.15 GB of data stored in MongoDB collections, respectively. We also propose a selection heuristic for effective identification of topics hidden in various data sources.
Authors: Neda Abolhassani (University of Georgia), Lakshmish Ramaswamy (University of Georgia),
Hide Authors & Abstract

Show Authors & Abstract

## SESSION 2B - Artificial Intelligence (FW-114., Faraday Wing Building) 11:00 - 13:00 ↓↑

Session Chair: Yu Weng
11:00 - 13:00
A next location predicting approach based on a recurrent neural network and self-attention

On most location-based social applications today, users are strongly encouraged to share activities by checking-in. In this way, vast amounts of user-generated da-ta can be accumulated, which include spatial and temporal information. Much re-search has been conducted on these data, which enables heightening the under-standing of human mobility. Therefore, the next location problem has attracted significant attention and has been extensively studied. In this paper, we propose a next location prediction approach based on a recurrent neural network and self-attention mechanism. Our model can explore sequence regularity and extract tem-poral feature according to historical trajectories information. We conduct our ex-periments on the location-based social network (LBSN) dataset, and the results indicate the effectiveness of our model when compared with the other three fre-quently-used methods.
Authors: Jun Zeng (School of Big Data & Software Engineering, Chongqing University, Chongqing, China), Xin He (School of Big Data & Software Engineering, Chongqing University, Chongqing, China), Ran Tang (School of Big Data & Software Engineering, Chongqing University, Chongqing, China), Hao Wen (School of Big Data & Software Engineering, Chongqing University, Chongqing, China),
Hide Authors & Abstract

Show Authors & Abstract
11:00 - 13:00
A Food Dish Image Generation Framework Based on Progressive Growing GANs

The generative adversarial networks (GANs) have demonstrated the ability to synthesize realistic images. However, there are few researches applying GANs into the field of food image synthesis. In this paper, we propose an extension to GANs for generating more realistic food dish images with rich detail, which adds a food condition that contains taste and other information. That makes the model generate images with rich details. To improve the quality of the generated image, the taste information condition is added to each stage of the generator and discriminator. First, the model learns embedding conditions of food information, including ingredients, cooking methods, tastes and cuisines. Secondly, the training model grows progressively, and the model learns details increasingly during the training process, which allows the model to generate images with rich details. To demonstrate the effectiveness of our proposed model, we collect a dataset called Food-121, which includes the names of the food, ingredients, cooking methods, tastes, and cuisines. The results of experiment show that our model can produce complex details of food dish image and obtain high inception score on the Food-121 dataset compared with other models.
Authors: Su Wang (Shanghai University), Honghao Gao (Shanghai University), Yonghua Zhu (Shanghai University), Weilin Zhang (Shanghai University), Yihai Chen (Shanghai University),
Hide Authors & Abstract

Show Authors & Abstract
11:00 - 13:00
Integration of Machine Learning Techniques as Auxiliary Diagnosis of Inherited Metabolic Disorders: Promising Experience with Newborn Screening Data

Tandem mass spectrometry is an advanced biochemical analysis method and has been widely used in screening of inherited metabolic disorders (IMDs). Obtained examination results are filtered by cutoff value and then interpreted based on doctor's knowledge to get diagnoses. However, cutoff-based approaches have difficulties with the correlations of multiple metabolites. Doctor's experiences affect the diagnostic decision-making as well. The rapidly increasing availability of newborn screening data (1.5M cases in this study) enables the application of machine learning (ML) techniques to provide more accurate diagnoses of IMDs compared to simple cutoff values. We investigated two tasks in this study, i.e. complicated patterns between metabolites and better auxiliary diagnostic means. Experimental results show that novel metabolic patterns found in the study are effective and meaningful. Integrating ML techniques with these patterns improved predictive performance compared to existing diagnostic methods, suggesting ML techniques are becoming valuable as auxiliary diagnostic tools.
Authors: Bo Lin (College of Computer Science and Technology, Zhejiang University), Jianwei Yin (College of Computer Science and Technology, Zhejiang University), Qiang Shu (The Children's Hospital, Zhejiang University School of Medicine), Shuiguang Deng (College of Computer Science and Technology, Zhejiang University), Ying Li (College of Computer Science and Technology, Zhejiang University), Pingping Jiang (The Children's Hospital, Zhejiang University School of Medicine), Rulai Yang (The Children's Hospital, Zhejiang University School of Medicine), Calton Pu (Department of Electrical and Computer Engineering, Georgia Institute of Technology),
Hide Authors & Abstract

Show Authors & Abstract
11:00 - 13:00
Nemesis: Detecting Algorithmically Generated Domains with An LSTM Language Model

Various malware families frequently apply Domain Generation Algorithms (DGAs) to generate numerous pseudorandom domain names to communicate with their Command and Control (C&C) servers. Security researchers make a lot of efforts to detect Algorithmically Generated Domains (AGDs) for fighting Botnets and relevant malicious network behaviors. In this paper, we propose a new AGD detection approach, Nemesis, based on a Long Short-Term Memory (LSTM) language model. Nemesis can identify whether given domain names are AGDs according to their string compositions, and without additional information. Nemesis first leverages an n-gram dictionary, which is built on real domain names, to tokenize domain names into n-grams. Then a pre-trained detector is used to classify domain names as real ones or AGDs according to the tokenized results. We evaluate Nemesis’ abilities to detect domain names generated by known DGAs and to discover new DGA families. It turns out that Nemesis can accurately detect AGDs with the precision of 98.6% and the recall of 96.7%. Besides, we verify that Nemesis largely outperforms several existing effective approaches.
Authors: Dunsheng Yuan (Institute of Information Engineering, Chinese Academy of Sciences), Ying Xiong (National Computer Network Emergency Response Technical Team/Coordination Center of China), Tianning Zang (Institute of Information Engineering, Chinese Academy of Sciences), Ji Huang (Institute of Information Engineering, Chinese Academy of Sciences),
Hide Authors & Abstract

Show Authors & Abstract
11:00 - 13:00
Positive-Unlabeled Learning for Sentiment Analysis with Adversarial Training

Sentiment classification is a critical task in sentiment analysis and other text mining applications. As a sub-problem of sentiment classification, positive and unlabeled learning or positive-unlabeled learning (PU learning) problem widely exists in real-world cases, but it has not been given enough attention. In this paper, we aim to solve PU learning problem under the framework of adversarial training and neural network. We propose a novel model for PU learning problem, which is based on adversarial training and attention-based long short-term memory (LSTM) network. In our model, we design a new adversarial training technique. We conducted extensive experiments on two real-world datasets. The experimental results demonstrate that our proposed model outperforms the compared methods, including the well-known traditional methods and state-of-the-art methods. We also report the training time, and discuss the sensitivity of our model to parameters.
Authors: Yueshen Xu (Xidian University), Lei Li (Xidian University), Jianbin Huang (Xidian University), Yuyu Yin (Hangzhou Dianzi University), Wei Shao (RMIT University), Zhida Mai (Xanten Guangdong Development Co., Ltd), Lei Hei (Xidian University),
Hide Authors & Abstract

Show Authors & Abstract
11:00 - 13:00
CNASV: A Convolutional Neural Architecture Search-Train Prototype for Computer Vision Task

Neural Architecture Search (NAS) has become more and more prevalent in the field of deep learning in the past two years. Existing work often focus on image classification, and few works recently extend NAS to another computer vision task, such as semantic image segmentation. The semantic image segmentation is essentially a dense prediction for each pixel on whole image. Therefore, we choose the same basic primitive operations to build the search space for the two computer vision task respectively. Searching good neural network architectures and then training them from scratch is a regular procedure for NAS. In this paper, we design a prototype system that deploy search module and train module to collaborate with each other. Follow the former work, we initialize over-parameterized cells architecture and then transform to the continuous relaxation of the architecture to derive the good subnetwork by gradient descent. Our system can support any differential search algorithm, such as one-shot, DARTS or ProxylessNAS. We illustrate the effectiveness of our chosen primitive operations in the image classification and ability to transfer these operations to build search space for semantic image segmentation.
Authors: Tianbao Zhou (College of Information Engineering, Minzu University of China), Yu Weng (College of Information Engineering, Minzu University of China), Guosheng Yang (College of Information Engineering, Minzu University of China),
Hide Authors & Abstract

Show Authors & Abstract
11:00 - 13:00
Web Services Classification with topical attention based Bi-LSTM

With the rapid growth of the number of Web services on the Internet, how to classify web services correctly and efficiently become more and more important in the development and application of Web services. Existing function-based service clustering techniques have some problems, such as the sparse document semantics, unconsidered word order and the context information, so the accuracy of service classification needs to be further improved. To address this problem, this paper exploits the attention mecha-nism to combine the local implicit state vector of Bi-LSTM and the global LDA topic vector, and proposes a method of Web services classification with topical attention based Bi-LSTM. Specifically, it uses Bi-LSTM to au-tomatically learn the feature representation of Web service. Then, it utilizes the offline training to obtain the topic vector of Web service document and performs the topic attention strengthening processing for Web service fea-ture representation, and obtains the importance or weight of the different words in Web service document. Finally, the enhanced Web service feature representation is used as the input of the softmax neural network layer to perform the classification prediction of Web service. The experimental re-sults validate the efficiency and effectiveness of the proposed method.
Authors: Yingcheng Cao (Hunan University of Science and Technology), Jianxun Liu (Hunan Univeristy of Science and Technology), Buqing Cao (Hunan Univeristy of Science and Technology), Min Shi (Hunan Univeristy of Science and Technology), Yiping Wen (Hunan Univeristy of Science and Technology), Zhenlian Peng (Hunan Univeristy of Science and Technology),
Hide Authors & Abstract

Show Authors & Abstract
11:00 - 13:00
Relation Extraction toward Patent Domain based on Keyword Strategy and Attention+Bi_LSTM Model

Patent terminology relation extraction is of great significance to the construc-tion of patent Knowledge graph. In order to solve the problem of long-distance dependency in traditional depth learning, a new method of patent terminology relation extraction is proposed, which combines attention mech-anism and bi-directional LSTM model and with keyword strategy. Category keyword features in each sentence obtained by the improved TextRank with the patent text information vectorization added. BiLSTM neural work and at-tention mechanism are employed to extract the temporal information and sentence-level global feature information. Moreover, pooling layer is added to obtain the local features of the text. Finally, we fuse the global features and local features, and output the final classification results through the softmax classifier. The addition of category keywords improves the distinc-tion of categories. Substantial experimental results demonstrate that the pro-posed model outperform the state-of-art neural model in patent terminology relation extraction
Authors: Xueqiang Lv (Beijing Key Laboratory of Internet Culture and Digital Dissemination Research, Beijing Information Science & Technology University), Xiangru Lv (Beijing Key Laboratory of Internet Culture and Digital Dissemination Research, Beijing Information Science & Technology University), Xindong You (Beijing Key Laboratory of Internet Culture and Digital Dissemination Research, Beijing Information Science & Technology University), Zhian Dong (Beijing Key Laboratory of Internet Culture and Digital Dissemination Research, Beijing Information Science & Technology University),
Hide Authors & Abstract

Show Authors & Abstract

## SESSION 3B - Software Development (FW-114., Faraday Wing Building) 13:30 - 14:30 ↓↑

Session Chair: Kuang Li
13:30 - 14:30
Predicting the Fixer of Software Bugs via a Collaborative Multiplex Network: Two Case Studies

Bug triaging is an essential activity of defect repair, which is closely related to the cost of software maintenance. Researchers have proposed automatic bug triaging approaches to recommend bug fixers more efficiently and accurately. In addition to text features, most of the previous studies focused on single-layer bug tossing (or reassignment) graphs, but they ignored the multiplex (or multi-layer) network characteristics of human cooperative behavior. In this study, we build a collaborative multiplex network composed of a tossing graph and an e-mail communication graph in the bug triaging process. By integrating the idea of network embedding and multiplex network measures, we propose a new strategy of random walks. Moreover, we present a bug fixer prediction model that takes structure and text features as inputs. Experimental results on two large-scale open-source projects show that the proposed method outperforms the selected baseline approaches in terms of commonly-used evaluation metrics.
Authors: Yutao Ma (School of Computer Science, Wuhan University), Jinxiao Huang (School of Computer Science, Wuhan University),
Hide Authors & Abstract

Show Authors & Abstract
13:30 - 14:30
Maintainable Software Solution Development using Collaboration between Architecture and Requirements in Heterogeneous IoT Paradigm

Internet of Things (IoT) has been tremendously involved in the development of smart infrastructure. Software solutions in IoT have to consider lack of abstractions, heterogeneity, multiple stakeholders, scalability, and interoperability among the devices. The developers need to implement application logic on multiple hardware platforms to satisfy the fundamental business goals. Moreover, long-term maintenance issues due to the frequent introduction of new requirements and hardware platforms pose a vital challenge in IoT solution development. Numerous techniques have been devised to satisfy the issues mentioned above for ubiquitous and smart infrastructure. However, these techniques lack in providing a comprehensive approach in dealing with the above challenges. In this paper, we argue that fundamentally, there is no difference between the architecturally significant requirements and the architectural design decisions in IoT solution development. The architecture revolves around the requirements gathered by the analyst at the requirements gathering phase. We stress that the requirements elicitation process must consider the software architectural assessment for maintainable software development. By adopting this perspective, we identify areas where both requirements and architecture communities collaborate to effectively increase the user acceptability, maintainability, and fulfill the heterogeneous needs of IoT solutions.
Authors: Wajid Rafique (Nanjing University), Wanchun Dou (Nanjing University), Maqbool Khan (Nanjing University),
Hide Authors & Abstract

Show Authors & Abstract
13:30 - 14:30
Type-based Modelling and Collaborative Programming for Control-Oriented Systems

Domain-specific languages are more expressive and therefore tackle complexity better, making software development easier and more efficient. They raise the level of abstraction and, together with domain-specific generators, can automate the creation of quality code. This paper proposes a type-base approach to requirement modelling, called CosRDL, to design high quality real-time embedded systems. A set of rules and formal methods are defined to build CosRDL models for embedded systems, from which the model may be verified apart the specification. CosRDL can specify the features of event-driven behaviors that support communication between active objects (processes) to support concurrency and collaborative computing. The control processing and properties can be described by CosRDL syntax as an model extension and to make system implementation model. Meanwhile, a case study is presented to illustrate our approach to requirement modelling of control systems.
Authors: weidong ma (Institute of Electronic Engineering, China Academy of Engineering Physics), Zhaohui Luo (Royal Holloway, University of London, UK),
Hide Authors & Abstract

Show Authors & Abstract
13:30 - 14:30
Predicting Traffic Flow Based on Encoder-Decoder Framework

Predicting traffic flow is of great importance to traffic management and public safety, and it has high requirements on accuracy and efficiency. However, the problem is very challenging because of high-dimensional features, spatial levels, and sequence dependencies. On the one hand, we propose an effective end-to-end model, called FedNet, to predict traffic flow of each region in a city. First, for the temporal trend, period, closeness properties, we obtain low-dimensional features by downsampling high-dimensional input features. Then we perform temporal fusion to get temporal aggregations of different spatial levels. Next, we generate traffic flow by upsampling the fused features which are obtained by combining the corresponding temporal aggregation and the output of the previous upsample block. Finally, the traffic flow is adjusted by external factors like weather and date. On the other hand, we transfer the original task into a sequence task and then use teacher forcing to train our model, which make it learn the sequence dependencies. We conduct extensive experiments on two types of traffic flow (new-flow/end-flow and inflow/outflow) in New York City and Beijing to demonstrate that the FedNet outperforms five well-known methods.
Authors: Xiaosen Zheng (Central South University), Zikun Yang (Central South University), Liwen Liu (Central South University), Li Kuang (Central South University),
Hide Authors & Abstract

Show Authors & Abstract

## SESSION 4B - Collaborative Applications for Recognition and Classification (FW-114., Faraday Wing Building) 14:30 - 15:30 ↓↑

Session Chair: Bin Cao
14:30 - 15:30
GeoCET: Accurate IP Geolocation via Constraint-based Elliptical Trajectories

The geographical location of the IP device is crucial for many network security applications, such as location-aware authentication, fraud prevention, and security-sensitive forensics. Since most data mining-based methods are subject to the privacy protection policies, the delay-based measurement methods have broader application prospects. However, these methodologies are relying on heavyweight traffic on networks and high deployment costs. Besides, the worst case errors in estimation made by delay-based measurement methods render them ineffective. In this paper, we propose an accurate IP geolocation approach called GeoCET. This methodology only requires a small number of one-way delays (OWDs) to locate the targets, combining with elliptical trajectory constraints and maximum log-likelihood estimation technique. We introduce polynomial regression to fit the delay-distance model and enhance the accuracy of the localization. To evaluate GeoCET, we leverage real-world data which come from China, India, Western United States, and Central Europe. Experimental results demonstrate that GeoCET performs better for all existing measurement-based IP geolocation methodologies.
Authors: Fei Du (School of Cyber Security, University of Chinese Academy of Sciences), Xiuguo Bao (National Internet Emergency Center, CNCERT/CC), Yongzheng Zhang (Institute of Information Engineering, Chinese Academy of Sciences), Huanhuan Yang (National Internet Emergency Center, CNCERT/CC),
Hide Authors & Abstract

Show Authors & Abstract
14:30 - 15:30
Covering Diversification and Fairness for Better Recommendation

Smart applications are appealing an accurate matching between users and items, in which recommendation technologies are applied widely. Since recommendation serve for two roles, namely users and items, accuracy is not the only focus, the diversification and fairness should also be paid more attention for improving recommendation performance. The tradeoff among the accuracy, diversification and fairness on recommendation is bringing a big challenge. This paper proposed a novelty recommendation model to ensure the recommendation performance, which introduces a multi-variate linear regression model to cooperate with the collaborative filtering method. This study utilizes an improved similarity metrics to discover the closeness between users and item categories under the help of the collaborative filtering methods, and exploits the micro attribute information of items by a multi-variate linear regression model to decide the final recommended items. The experimental results show that our proposed method can provide better recommendation accuracy, diversification and fairness than the recommendation based on pure collaborative filtering method.
Authors: Qing Yang (Guilin University of Electronic Technology), Li Han (Guilin University of Electronic Technology), Ya Zhou (Guilin University of Electronic Technology), Shaobing Liu (Guilin University of Electronic Technology), Jingwei Zhang (Guilin University of Electronic Technology), Zhongqin Bi (Shanghai University of Electronic Power), Fang Pan (Guangxi Normal University),
Hide Authors & Abstract

Show Authors & Abstract
14:30 - 15:30
Towards Efficient Pairwise Ranking for Service Using Multidimensional Classification

With the growing popularity of services which meet the divergent requirements from users, service selection and recommendation have drawn significant attention in services computing community. Service ranking is the most important part in service selection and recommendation. Although there have been several existing approaches of service ranking which is basically rating-based, suffering from the heterogeneity of ranking criteria from users. Moreover, the efficiency of such comparison-based approaches is the bottleneck in reality. To attack these challenges, an efficient pairwise ranking scheme with multidimensional classification is proposed in this paper, which also fully considers the context information of service and users. Furthermore, the scheme is able to mitigate data sparsity of users similarity matrix and improve accuracy. Next, we introduce a random walk model for ranking formulation, and propose a Markov chain based approach to obtain the global ranking. Finally, the efficacy of our approach is validated by experiments adopting the real-world YELP dataset.
Authors: Yingying Yuan (Beijing University of Posts and Telecommunications), Jiwei Huang (China University of Petroleum - Beijing), Yeping Zhu (Chinese Academy of Agricultural Sciences), Yufei Hu (Beijing Boyu Kaixin Machinery Equipment Co., Ltd.),
Hide Authors & Abstract

Show Authors & Abstract
14:30 - 15:30
Collaborative Computing of Urban Built-up Area Identification from Remote Sensing Image

Urban built-up area is one of the important criterions of urbanization. Remote sensing can quickly acquire dynamic temporal and spatial variation of urban built-up area, but how to identify and extract urban built-up area information from massive remote sensing data has become a bottleneck arousing widespread con-cerns in the field of the data mining and application for remote sensing. Based on the traditional urban built-up area identification and data mining of remote sens-ing, this paper proposed a new collaborative computing method for urban built-up area identification from remote sensing image. In the method, the normalized difference built-up index (NDBI) and the normalized differential vegetation index (NDVI) feature images were constructed firstly from the spectrum clustering map; and then the urban built-up area was identified and extracted by the map-spectrum synergy and mathematical morphology methods. Finally, a case of col-laborative computing of urban built-up areas in Chongqing city, China is present-ed. And the experimental results show that the total accuracy of urban built-up ar-ea identification in 1988 and 2007 reached 92.58% and 91.41%, the Kappa coef-ficient reached 0.8933 and 0.8722, respectively, and the good results in the tem-poral and spatial variation monitoring of urban built-up area are achieved.
Authors: chengfan li (Shanghai University), lan liu (Shanghai University of Engineering Science), yongmei lei (Shanghai University), xiankun sun (Shanghai University of Engineering Science), junjuan zhao (Shanghai University),
Hide Authors & Abstract

Show Authors & Abstract
Day 4 22/08/2019
Room #1

## SESSION 5 - Smart Transportation (Nelson Haden Lecture Theatre, Faraday Wing Building)） 07:00 - 08:00 ↓↑

Session Chair: Liu Jing
07:00 - 08:00
Intelligent-prediction Model of Safety-risk for CBTC System by Deep Neural Network

Safety-risk estimation aims to provide guidance of the train's safe operation for communication-based train control system (CBTC) system, which is vital for hazards avoiding. In this paper, we present a novel intelligent-prediction model of safety-risk for CBTC system to predict which kind of risk state will happen under a certain operation condition. This model takes advantages of popular deep learning models, which is Deep Belief Networks (DBN). Some risk prediction factors is selected at first, and a critical function factor in CBTC system is generated by statistical model checking. Afterwards, for each input of samples, the model utilizes DBN to extract more condensed features, followed by a softmax layer to decouple the features further into different risk state. Through experiments on real-world dataset, we prove that our new proposed intelligent-prediction model outperforms traditional methods and demonstrate the effectiveness of the model in the safety-risk estimation for CBTC system.
Authors: Yan Zhang (Shanghai Key Laboratory of Trustworthy Computing, East China Normal University, Shanghai, China), Jing Liu (Shanghai Key Laboratory of Trustworthy Computing, East China Normal University, Shanghai, China), Junfeng Sun (R&D Institute, CASCO Signal Ltd., Shanghai, China), Xiang Chen (R&D Institute, CASCO Signal Ltd., Shanghai, China), Tingliang Zhou (R&D Institute, CASCO Signal Ltd., Shanghai, China),
Hide Authors & Abstract

Show Authors & Abstract
07:00 - 08:00
A Platform Service for Passenger Volume Analysis on Massive Smart Card Data in Public Transportation Domain

In current public transportation of modern cities, the passenger volume analysis counts the bus passengers in multiple perspectives, and it is signif-icant to optimize the bus scheduling and evaluate transportation capacity. On the smart card data of passengers taking buses, traditional solutions have inherent limitations about long processing delay, inaccuracy result and poor scalability. In this paper, the spatio-temporal correlation with business restrictions is considered, and an effective platform service for passenger volumes analyses are proposed on massive smart card. Our service has been applied in practical usage for three types of passenger volume, and holds minute-level latencies on weekly data with nearly linear scalability in extensive conditions.
Authors: Weilong Ding (North China University of Technology), Zhe Wang (North China University of Technology), Zhuofeng Zhao (North China University of Technology),
Hide Authors & Abstract

Show Authors & Abstract
07:00 - 08:00
Reliable Collaborative Semi-Infrastructure Vehicle-to-Vehicle Communication for Local File Sharing

Recently, Vehicular Cloud Communication (VCC) has been gaining momentum targeting intelligent and efficient data transmission. VCC is a type of mobile ad-hoc network comprising heterogeneous vehicles sharing their re-sources to perform collaborative activities. In this paper, we propose a new semi-infrastructure file-browsing in order to provide Network as a service (NaaS) enabling internet-independent browsing. In our scenario, a central management platform plays the role of controlling and managing the selection of re-laying vehicles supporting the source to destination file transmission procedure. Nagel-Schreckenberg rules for traffic cellular automata (CA) are used as the basis for our scenario simulation. Nagel-Schreckenberg rules simulate the behavior of a group of hypothetical vehicles moving across a highway. We study the reliability and efficiency of file transfer in such settings. Simulation results show that the number of selected relays required to establish the network highly impacts the probability of successfully sending the requested files. In addition, the distances between the selected relays influence the network throughput and the probability of network failure. Moreover, the density of relays strongly affects the overall delay that occurs due to the continuous re-transmission of the selected files among different hops.
Authors: Bassem Mokhtar (Electrical Engineering Department, Alexandria University, Alexandria, Egypt.), Mohamed Azab (Department of Computer and Information Sciences, Virginia Military Institute, USA.), Efat Fathalla (Department of Electrical and Computer Engineering, Old Dominion University, Norfolk, VA, USA), Esraa M. Ghourab (Electrical Engineering Department, Alexandria University, Alexandria, Egypt.), Mohamed Magdy (College of computing and information technology, Arab Academy for Science, Technology and Maritime Transport, Alexandria, Egypt), Mohamed Eltoweissy (Department of Computer and Information Sciences, Virginia Military Institute, USA.),
Hide Authors & Abstract

Show Authors & Abstract
07:00 - 08:00
An Efficient Mutual Authentication Framework with Conditional Privacy Protection in VANET

Vehicular Ad Hoc Network ( VANET ) is a special application of traditional Mobile Ad Hoc Network ( MANET ) in traffic roads, which has attracted extensive attention due to its important role in intelligent traffic and road services. In order to ensure the safety of road traffic and protect the privacy of users, it is of vital importance to provide effective anonymous authentication in VANET. In this paper, we propose an efficient mutual authentication framework with conditional privacy protection(EMAPP), which can achieve the security authentication from vehicles to infrastructure and vehicles to vehicles. In the proposed framework, we are combined with pseudo ID and temporary pseudonym to protect the privacy of vehicles, and use the identity-based signature scheme to achieve authentication between vehicles and infrastructure. At the same time, with the assistance of the roadside unit(RSU), we utilize an online/offline signature scheme to achieve authentication between vehicles in the same RSU area and different RSU area. Our scheme has reusability, and we have conducted a performance evaluation. Without expensive and time-consuming operations such as bilinear pairing and mapping to point(MTP) functions, our framework can produce better performance and is appropriate for practical application. In addition, we also use the Internet Security Protocol and Application Automatic Authentication (AVISPA) tools to provide formal security analysis.
Authors: Ying Wang (Tianjin University), Jing Hu (Tianjin University), Xiaohong Li (Tianjin University), Zhiyong Feng (Tianjin University),
Hide Authors & Abstract

Show Authors & Abstract

## SESSION 6 - Workshops SITN, WCCA (Nelson Haden Lecture Theatre, Faraday Wing Building) 08:30 - 10:00 ↓↑

Session Chair: Thomas Tan & Ying Zhang
08:30 - 10:00
¬Lightweight Computation to Robust Cloud Infrastructure for Future Technologies

Recently, lightweight technologies has gain a lot of attentions for future emerging technologies and play a vital role in smart cities. As, it is providing more efficient and fastest solution than traditional virtual machines and capable to serves in heterogeneous and dynamic environment across multiple domains including IoT, Fog computing and multi-access edge computing. In this paper, we proposed a light-weight solution for LCC (Live Container Cloud) that permits the user to access live/remote cloud resources within container. LCC can be embed as a fog/ edge node to permit the users to access, allocate and deallocate cloud resources. Moreover, the effectiveness of container technology is presented in terms of its performance
Authors: Sonia Shahzadi (Swan Mesh Networks Ltd, Research and Development, London, UK.), Muddesar Iqbal (London South Bank University, London, United Kingdom,), George Ubakanma (London South Bank University, London, United Kingdom), Tasos Dagiuklas (London South Bank University, London, United Kingdom), Andrei Andrei Tchernykh (CICESE Research Center, Ensenada, Baja California, Mexico),
Hide Authors & Abstract

Show Authors & Abstract
08:30 - 10:00
Evaluation of Underlying Switching Mechanism for Future Networks with P4 and SDN

Future networks and applications requires highest possible bandwidth and lowest congestion in order to maximize the customer satisfaction and achieve requirements for heterogeneous applications. With this motivation in mind Software Defined Networking (SDN) was introduced which was first released, targeting data centers and local area networks. At present SDN covers all aspects of the network covering multiple scenarios and use cases. Programming Protocol independent Packet Processing (P4) is a high level language capable of forwarding packets via a programmable parser followed by multiple stages of match + action which is considered the most effective mechanism for routing in IP networks. This paper takes into the account the latest platforms developed for service providers, Open Networking Operating System (ONOS) to deploy two environments configured in the aforementioned technologies in order to test their performance. Two case studies were drawn with four experiments simulating in Mininet for each, to further understand and evaluate the performance of aforementioned technologies for switching. By incorporating SDN + P4 environment a significant reduction of 92% and 70% was achieved in destination host discovery and UDP packet transmission delay respectively. E2E was reduced by 7% by SDN + P4 case study compared to the SDN only environment.
Authors: Omesh Fernando (University of Hertfordshire), Hannan Xiao (University of Hertfordshire), Xianhui Che (University of Hertfordshire),
Hide Authors & Abstract

Show Authors & Abstract
08:30 - 10:00
A Novel Feature-selection Approach Based on Particle Swarm Optimization Algorithm for Intrusion Detection Systems

Aiming at the KDDCUP99 data set, studying the selection of intrusion detection features. An intrusion detection feature selection method based on discrete particle swarm optimization (PSO) is proposed. Based on the correlation evaluation method, according to the correlation of the particles, 14 main feature attributes were selected from all 41 features of the KDDCUP99 data set as intrusion detection features. 57141 data were extracted from the 10% KDDCUP99 data set to construct the experimental sample space, and the classification model was built using the classic classifier built into the Weka platform of the intelligent analysis environment. The ten-fold cross-validation method was used for training and verification. The results show that the proposed feature selection algorithm is applicable to a variety of classifiers, which not only improves the correct classification ratio, but also shortens the model time of the classifier.
Authors: Jianzhen Wang (Business College of Shanxi University), Yan Jin (Business College of Shanxi University),
Hide Authors & Abstract

Show Authors & Abstract
08:30 - 10:00
Multiuser Detection using Hybrid ARQ with Incremental Redundancy in Overloaded MIMO Systems

Multiple Input and Multiple Output (MIMO) systems use multiple antennas at both transmitter and receiver end for increasing link capacity and spectral efficiency. However,combining schemes used for such systems face critical issues such as presence of interference, existence of multiple signals to interference and noise power ratios (SINRs) and complexity. To overcome the asserted issues; in this paper linear multiuser detection techniques are employed in over loaded MIMO systems where the number of transmit antennas (Nt) is greater than number of receiver antennas (Nr), using Hybrid Automatic Repeat request with Incremental Redundancy (HARQ-IR). The primary aim of this research is to enhance bit error rate (BER) and throughput by transforming an overloaded MIMO systems (Nt > Nr) into critically loaded system (Nt = Nr) or under loaded MIMO systems (Nt < Nr) by simple retransmission method. Simulation results show unprecedented performance compared to contemporary approaches in term of throughput and BER.
Authors: Muhammad Kashif (Department of Electrical and Computer Engineering Center for Advanced Studies in Engineering (CASE) Islamabad, Pakistan), Zakir Ullah (Department of Electrical and Computer Engineering Center for Advanced Studies in Engineering (CASE) Islamabad, Pakistan), Muddesar Iqbal (London South Bank University, London, United Kingdom,), Leila Musavian (School of Computer Science and Electronic Engineering University of Essex UK), Xinheng Wang (School of Computing and Engineering University of West London UK), Shahid Mumtaz (Instituto de Telecomunicaes Portugal), Sohail Sarwar (National University of Sciences and Technology (NUST) Rawalpindi, Pakistan), Zia ul Qayyum (Department of Computer Science Allama Iqbal Open University Islamabad, Pakistan), M Safyan (Department of Computer Science GC University Lahore Lahore, Pakistan),
Hide Authors & Abstract

Show Authors & Abstract
08:30 - 10:00
Quantum Based Networks: Analysis of Quantum Teleportation Protocol and Entanglement Swapping

In this paper we consider the quantum teleportation and entanglement swapping protocols used in quantum based networks for passing information between a sender and receiver. For the teleportation protocol we observe and identify relationships that exist between Einstein-Podolsky-Rosen (EPR) Bell states employed as quantum resources, measured sender values and the gates employed at the receiver side. For the entanglement swapping protocol we consider input and output EPR states and the relationship between the two. We include a review of the concepts and our findings from the analysis carried out.
Authors: Preeti Kandwal (University of Hertfordshire, Hatfield, United Kingdom), William Spring (University of Hertfordshire, Hatfield, United Kingdom), Hannan Xiao (University of Hertfordshire, Hatfield, United Kingdom),
Hide Authors & Abstract

Show Authors & Abstract
08:30 - 10:00
A Decentralized and Anonymous Data Transaction Scheme Based on Blockchain and Zero-knowledge Proof in Vehicle Networking

Data transaction in internet of vehicles is a transaction occurs between vehicle owner and data buyer. Blockchain is a new technology that brings decentralized ledger system for user, which means users could make payment without the third party. There are several projects combined internet of vehicles and Blockchain, however, none of them realize a trustworthy anonymous data transaction. In this paper, we first propose the concept of Super Nodes to guarantee data authenticity, then we construct the anonymity for the transaction base on zero-knowledge Succinct Non-interactive Argument of knowledge (zk-SNARKs) and DAP from Zerocash. Moreover, a smart contract is deployed for mutual benefits. Simulation experiment shows this scheme is practical.
Authors: Wei Ou (Hunan University of Science and Engineering, Yongzhou, China),
Hide Authors & Abstract

Show Authors & Abstract