Applied informatics research

Australian Research Council (ARC) funding

Effective and Efficient Situation Awareness in Big Social Media Data. ARC Discovery project, $510K, 2020-2022. Preventing sensitive data exfiltration from insiders. ARC Discovery project, $600K, 2020-2022.

Increasing data quality with group associations in outsourcing environments. ARC Discovery project. $378,000, 2018-2020.

Deep mining neurological abnormalities from brain signal data. ARC linkage project. $610,000, 2018-2021.

Privacy Preserving Data Sharing in Electronic Health Environment. ARC Linkage Project. $540,000, 2015-2019.

Australia Government & Industry Grants

Efficient and Secure Cloud - Based Healthcare Systems for the Storage of Electrical Medical Records, OCSC project, 48K, 2018-2019.

Detection and prevention of data exfiltration by insiders in database systems, Germany – Australia research funding, 30K, 2018-2019.

Abnormality detection in banking environments

Due to widespread adoption of electronic funds transfer at point of sale (EFTPOS), internet banking and the near-ubiquitous use of credit cards, banks are able to collect abundant information about cardholders’ transactions. Extracting effective knowledge from these transaction records and personal data from cardholders has enormous profit potential for the banking industry.

In this project, we provide a novel domain-driven classification method that takes advantage of multiple criteria and multiple constraint-level programming (MC2) for intelligent credit scoring. The proposed method will classify credit card customers precisely in order to provide effective services while avoiding losses due to users’ unpaid debts.

This project was supported by a VU grant (2013).

Research team:

• Professor Yanchun Zhang, Director PAI, VU
• Dr Xin Wang, ANZ Bank
• Dr Mehmet Yildiz, IBM Australia
• Professor Yong Shi, Chinese Academy of Sciences

Event detection based on Twitter text analysis

Event detection on social media site Twitter has become a promising research direction due to Twitter’s popularity, up-to-date features, free writing style and so on. Unfortunately, it’s a challenge to analyse Twitter dataset for event detection, since the informal expressions of short messages comprise many abbreviations, Internet buzzwords, spelling mistakes and meaningless content.

Previous techniques proposed for Twitter event detection mainly focus on clustering words related to the events, while ignoring that these words may not be easily interpreted to clear event descriptions. This project will investigate the tasks of aggregation of tweets to reveal the collective intelligence; the spatiotemporal identification of events (such as public health events); and analysis of event trends.

This project was supported by the Australia-China Science and Research Fund (ACSRF), Australia Government grant (2013).

Text mining & web search

This project aims to provide an effective health vertical search engine based on domain-specific text mining of web pages. This search engine will return fewer but more exact hits than a general search engine and allow users to input better health-related terms for queries through a hint tool. The project outcome will include a health vertical search engine system, a set of Application Programming Interfaces (APIs) for accessing the search engine database, and several novel text mining approaches.

This project was supported by a VU grant (2013).

Researcher: Professor Yanchun Zhang, Director PAI, VU

Edutainment

Learning is a major activity for most children aged 3 to 22 years old, and is therefore a key period in human life. Researchers in Education always strive to make learning an enjoyable activity.

Game-based learning has a long history. However, classic game-based learning requires a lot of extra effort and resources (including human resources), thus is only applied in a limited way in very early-year education. The current widespread usage of computer games provides a great opportunity to design edutainment systems for kids to gain knowledge and skills in a fun way.

In our experiments of embedding learning content in virtual world games have shown its significant power in retaining students learning interests and improving their performance. In edutainment, there are two types of knowledge to be modelled: the knowledge to be learned and the knowledge on how learning should be facilitated. This project was supported by a VU grant (2008-2009).

Researcher: Professor Yuan Miao

Learning through negotiation

Argumentation plays an important role in promoting deep learning, fostering conceptual change and supporting problem solving. The new “learning by arguing” paradigm leads to new learning opportunities. However, due to the difficulties in modelling human cognition, there are few learning systems that can facilitate argumentation dialogues between systems and learners.

Towards the goal of providing an easy to use effective knowledge model to facilitate learning through (automatic) negotiation, we apply a number of computational models including cognitive maps and classic rule based systems. Particularly, cognitive maps are a family of computational models for capturing human knowledge and facilitating machine inferences. They have gained increasing popularity among domain applications because CMs are easy to use, can easily model domain experts’ knowledge, have a visualised presentation of the modelled knowledge, and have a clear mapping between vertices in the model and the corresponding factors in the real system.

Based on these knowledge models we design intelligent software agents to facilitate argumentative learning. The agents are able to simulate a peer learner and automatically conduct argumentative dialogues with learners. All knowledge based interaction can be viewed as a form of learning process, or an argumentation based learning. The argumentative agents can be applied in general school education as well as special domains like diabetes education and eHealth decision support.

Researcher: Professor Yuan Miao

Water resource management requires access to meteorological data, spatial data such as digital elevation, land use and remote sensing data. Current approaches to water resource management, for the most part, are operating independently while generally working towards the same objective (for example, improved water quality in waterways in a region).

While each of these activities has been effective in their own right, there are major benefits and synergies to be gained through a coordinated integration of these tasks. Therefore, it is crucial and promising to deliver research on water resource management and Information and Communication Technologies related issues, including:

• the framework for the integration of all data and modelling services related to water resource management for a region
• field monitoring with its associated spatial temporal trend analysis
• remote sensing (including satellite and autonomous stations)
• predictive modelling that facilitates and supports management decisions.

Semantics Water Web based on spatial temporal data mining

Adaptive water management requires the assimilation of many different, varied data sets in order to obtain a holistic ("whole-of-water-cycle") view of the environment. The development of Semantics Water Web is used for rapid, seamless integration of the many independently developed water data sources and models based on:

• spatial temporal data mining
• wireless sensor networks
• web services with novel algorithms and computational techniques for the successful analysis of large spatial-temporal water quality databases and the disclosure of interesting knowledge on remote sensing
• geographical information systems
• computer cartography
• environmental assessment
• water planning issues underpinning decision makings.

Research team:

• Professor Yanchun Zhang, Director PAI, VU
• Dr Guangyan Huang, VU
• Professor Stephen Gray, VU
• Professor Yong Shi, Chinese Academy of Sciences

Spatial-temporal data mining for water resource decision support

The aim of this project is to construct spatial data mining system for water resource decision support based on geospatial and temporal database by improving the quality, completeness, relevance and interpretability of the water data, and building the appropriate spatial data mining model underpinning these decisions.

Spatial data carries topological and/or distance information and it is often organised by spatial indexing structures and accessed by spatial access methods. These distinct features of a spatial database pose challenges and bring opportunities for mining information from spatial data. In this study the spatial and temporal database has been mined using a series of innovative optimisation-based programming algorithm on remote sensing, geographical information system, computer cartography, environmental assessment and planning issues. A further national benefit will be the development of frontier spatial data mining techniques to ensure Australia a leading role in data engineering and knowledge discovery services.

Research team:

• Professor Yanchun Zhang, Director PAI, VU
• Professor Stephen Gray, VU
• Professor Yong Shi, Chinese Academy of Sciences

Data Enhancement, integration & access services for smarter, collaborative & adaptive whole-of-water cycle management

The aim of this project is to improve the speed, rigour and adaptability of the decisions made within South East Queensland by the partners in the Healthy Waterways Partnership by focussing on services that will improve the quality, completeness, relevance and interpretability of the data used in the models underpinning these decisions.

The project is expected to contribute to improved water quality and healthier ecosystems. It is supported by ARC Linkage Project with South East Queensland Healthy Waterways Partnership as an industry partner.

Research team:

• Professor Yanchun Zhang, Director PAI, VU
• Professor Xiaofang Zhou, University of Queensland
• Professor Jane Hunter, University of Queensland
• Associate Professor Shazia Sadiq, University of Queensland
• Dr Eva Abal, South East Queensland Healthy Waterways

E-health is playing an increasingly important role in healthcare. E-health refers to the use of information and communication technologies (ICTs) in the health domain. E-health has diverse applications, such as:

• health information networks
• telemedicine
• electronic medical records
• health knowledge management
• personal wearable and portable communicable systems
• health portals
• health research data.

ICT-supported information platforms will improve decision-making processes and quality management in healthcare. The research of e-health will explore a range of practical strategies to enable ICTs to be integrated into health and wellbeing projects, with particular focus on risk detection and prevention, and health data management. Some of our current e-Health projects are below.

WHO eSTEPS 2.0

In partnership with the World Health Organization (WHO), the eSTEPS 2.0 software is a handheld computer-based data collection system. It aims to transform the paper-based STEPS questionnaire into an electronic form that can run from a PDA, incorporating automatic skips, valid range checks and automatic assigning of ID numbers, and other functions for data collections.

Research team:

• Professor Yanchun Zhang,  VU
• Associate Professor Hao Shi, VU
• Professor Yuan Miao, VU
• Khai Hoang, MSc student, VU
• Yanan Hao, PhD student, VU
• Associate Professor Maximilian de Courten, Monash University
• Leanne Riley, WHO
• Melanie Cowan, WHO

Tasmanian web-based epidemiology system

The Tasmanian web-based epidemiology system is software that taps into Google Maps technology to identify healthcare needs by geographical location. It has enabled the Tasmanian Department of Health Services to better use its hospitalisation, death, cancer incidence and diseases data for monitoring and surveillance of the health of Tasmanian residents.

Research team:

• Professor Yanchun Zhang, Director PAI, VU
• Associate Professor Hao Shi, VU
• Jingyuan Zhang, PhD student, VU
• Dr Peter Wan, DHHS, Tasmania
• Dr Kelly Shaw, DHHS, Tasmania

GP-eConnect

In partnership with the Australian Research Council and Westgate General Practice Network, this project aims to vastly improve management of medical records by allowing referrals and patient data to be integrated for use by different service providers.

We developed a GP E-connect system to facilitate medical data exchange and service integration over different medical systems. It significantly improves the efficiency and quality of services to patients while providing automatic routine processes and easily being used by general practitioners, specialists, pharmacists and hospitals, with lower cost of system maintenance.

Research team:

• Professor Yuan Miao, PAI, VU
• Professor Yanchun Zhang, PAI, VU
• Dr Xuehong Tao, Research Officer, VU
• Xuebing Yang, PhD student, VU
• Corinne Siebel, WGPN
• Manfred Queteschiner, WGPN

PhysAnalyser

In partnership with the Royal Brisbane & Women’s Hospital and the Australian Research Council, the PhysAnalyser software is being developed that allows surgeons and anaesthetists to monitor and predict risks to patients during surgery by comparing their past health records with immediate health-related information, such as current blood pressure and heart rate. It can also compare a patient’s data to other patients’ to predict and prevent life-threatening incidents.;

Research team:

• Professor Yanchun Zhang, Director PAI, VU
• Dr Guangyan Huang
• Peng Zhang, PhD student
• Dr Xun Yi
• Professor Michael Steyn, Royal Brisbane & Women's Hospital
• Professor Kersi Taraporewalla, Royal Brisbane & Women's Hospital
• Professor Jie Cao, Nanjing University of Finance and Economics, China

Protect information sharing within distributed collaborative environment

Information sharing on distributed collaboration usually occurs in broad, highly dynamic network-based environments, and formally accessing the resources in a secure manner poses a difficult and vital challenge. This project develops a systematic methodology for information sharing in distributed collaborative environments. It will ensure sensitive information and information assurance requirements, and incorporate new security constrains and policies raised by emerging technologies. We will create a new rule-based framework to identify and address issues of sharing in collaborative environments, and to specify and enforce security rules to support identified issues while minimising the risks of information sharing through the framework.

This project was supported by Australian Research Council (ARC) Discover Project (DP0988465).

Researcher: Professor Hua Wang, PAI, VU

Privacy preserving data sharing in data mining environments

Preserving privacy in data mining among various enterprises and organisations is essential for many real world applications in areas like health surveillance, business analysis, fraud detection and terror protection. Efficient and effective techniques are badly needed to protect privacy in data sharing and data mining. The developed cutting-edge techniques in this project will be implemented in freely available open source software tools, empowering Australian organisations to utilise the techniques to develop intelligent systems in data sharing environments. These techniques will ultimately lead to better utilisation of the information available in many enterprises and organisations.

This project was supported by Australian Research Council (ARC) Discover Project (DP0663414).

Research team:

• Professor Jiuyong Li, UniSA
• Professor Hua Wang, PAI, VU

Limiting disclosure of private information in relational database systems

Enterprises are deeply concerned about customers' privacy issues and try to build solid trust to attract customers. This project continues development of new purpose-based frameworks and private information assurance requirements in relational database systems. The frameworks will identify and address issues of protecting private information; and to specify and enforce privacy rules to support identified issues. It aims to develop techniques for purpose-based usage control and detecting possible conflicts between obligations. The approach leads to a great understanding of advocating limited disclosure in usage control systems. The project develops fundamental enabling methodologies for the information and communication industry.

This project was supported by Australian Research Council (ARC) Discover Project (DP0988465).

Researcher: Professor Hua Wang, PAI, VU

Privacy protection in distributed data mining & data warehouse query

Information and Communications Technology (ICT) has dramatically altered the world's social and economic landscape. 'From data to knowledge' is one of the priority challenges recognised by National ICT Australia. However, privacy concerns may prevent it from realisation. This project aims to fulfil 'from data to knowledge' without breaching privacy of data from distributed resources held by different parties. The outcomes of this project will create new directions in the research of privacy‑preserving distributed data mining and are applicable to Australian counter‑terrorism and homeland defence in detecting bio‑terrorism from privacy sensitive data.

This project was supported by Australian Research Council (ARC) Discover Project (2007-2009) and ARC Discover Project (2009-2011).

Research team:

• Professor Hua Wang, PAI, VU
• Professor Yanchun Zhang, Director PAI, VU
• Professor Eiji Okamoto, University of Tsukuba

Private searching on streaming data

Private searching on streaming data is a process to dispatch a program to a public server. The program searches streaming sources of data without revealing searching criteria and then returns a buffer containing the findings.

From an Abelian group homomorphic encryption, the searching criteria can be constructed by only simple combinations of keywords, e.g., disjunction of keywords. The recent breakthrough in fully homomorphic encryption has allowed us to construct arbitrary searching criteria theoretically. In this research, we consider a (t,n) threshold query, which searches for documents containing more than t out of n keywords. This form of query can help us find more relevant documents.

We have constructed the searching criteria for private threshold searching on streaming data on the basis of the state-of-the-art fully homomorphic encryption techniques. Our protocol is semantically secure as long as the underlying fully homomorphic encryption scheme is semantically secure.

Researcher: Professor Hua Wang, PAI, VU

Private protection for location-based queries

In this research, we have given a solution to one of the location-based query problems.

This problem is defined as follows:

1. a user wants to query a database of location data, known as Points Of Interest (POI), and does not want to reveal his/her location to the server due to privacy concerns.
2. the owner of the location data, that is, the location server, does not want to simply distribute its data to all users.

The location server desires to have some control over its data, since the data is its asset. Previous solutions have used a trusted anonymiser to address privacy, but introduced the impracticality of trusting a third party. More recent solutions have used homomorphic encryption to remove this weakness. Briefly, the user submits his/her encrypted coordinates to the server and the server would determine the user's location homomorphically, and then the user would acquire the corresponding record using Private Information Retrieval techniques.

We have proposed a major enhancement upon this result by introducing a similar two stage approach, where the homomorphic comparison step is replaced with Oblivious Transfer to achieve a more secure solution for both parties. Our solution is efficient and practical in many scenarios.

Researcher: Professor Hua Wang, PAI, VU