Thanh Hai Nguyen

My primary research now focuses on Automation Web Services Composition in Web Semantics. I am developing a completed end-to-end AI system to collect requirements from users in Natural Language and explore workflows that can satisfy the users requirements automatically. After the workflow is achieved, our system is able to execute each Web Service component in workflow sequence in order to achieve the goal. This system includes many parts in Artificial Intelligence area. They are :

• A Web Service Registry : An CMS (Content Management System) web-based system (PHP, MySQL, Drupal, HTML and JavaScript) is used to collect and store Web Services from service providers. It provides ways how to invoke and execute a web service. In AI, it is responsible about syntax of Web Service execution.
• An Ontology : An Ontology (OWL) is an explicit specication of a conceptualization. For AI system, what "exits" is that which can be represented. When the knowledge of a domain is represented in a declarative formalism, the set of objects that can be represented is called the universe of discourse. Ontology is responsible about semantics of Web Service execution.
• An AI Core Planning : It includes Planning Engine (Answer Set Programming and C language), Planning Monitor (C language) and Workflow Execution (Java) and Re-Composition Engine (Answer Set Programming and C). Planning Engine is a computer software to generate plan/workflow based on Web Service compostion algorithms and AI planning theory. Planning Monitor is responsible to monitor and control generation processes and provide additional functions such as failure correctness, adaptability, scalability, reliability. And Workflow Execution is a software that will execute and invoke all web service components in workflow/plan to achieve the Goal. Last component, Re-Composition Engine is responsible to re-generate updated workflows which contain additional users requirements and conditions as constraints - users change from current workflow. In which, "preserve as much as possible generated workflow" is implemented based on similarity index measurement. The new updated workflow with highest similarity index with current workflow will be considered to present to users. The similarity index measurement is developed by using Machine Learning techniques such as Ontology Semantic Comparison, BPM similarity, Graphs Topological Comparison, etc. integrated with Natural Language Processing (NLTK- Stanford CoreAPI).
• An Ontology Accessing Module: It is an software created based on SPARQL engine and programming languages (C and JAVA) in order to provide functionality and methods to access, retrieve and interact with Ontology classes, instances and properties. This software is a back-end of some components:
  • Ontology Web Service API: developed based on Python and provide serives API(s) to retrieve and access Ontology assets via Internet. It works with other UI components and Natural Language processing module to create end-users environment.
  • Translation program: developed based on C and Java in order to translate Ontology OWL language to any other programming languages such as Logic Programming, ASP, C, etc. These products will contribute with Planning Engine.
• Workflow Description Graphical Tool (WDGL) : This is an UI computer software used by end-users in order to provide tools to build an workflow based on user's requirements. Normal users will be provided enough tools to describle which input/output of composition, which services they want or don't want, generate workflow in visulization, change or modify origin workflow and re-generate new workflow.
• Natural Language Processing module : This module is created to describe workflow as human readable language and make Input/Output description to become much easier when end-users are not expert.

More detail information about this project.

• Workflow Configuration Module [Main Link][Supported Link]

Publications.

• PADL Paper

• ICLP Paper (TPLP Journal)

"Phylotastic" refers to a community-based system for delivering phylogenetic knowledge via web services composition. Individual web services provide taxonomy resolutions, source trees, pruning operations, calibrations, and so on. Clients of the system combine these services to provide species trees to users. The idea emerged from NESCent's HIP (Hackathons, Interoperability, Phylogenies) working group, and was prototyped at a NESCent-sponsored hackathon in 2012, as described by Stoltzfus et al, 2013.

More detail information about this project.

Main web site : http://phylotastic.org/

Portal Site: http://phylotastic.net

Mobile app: Apple App Store

Github repo :

"The Land-Potential Knowledge System (LandPKS) aims to serve as the primary access and development point for free, simple to use, and locally appropriate technologies and knowledge needed for making sustainable land management decisions. The program develops innovative web-based and mobile data collection and analysis methods and tools to support local land use planning and to optimize design and implementation of food security, land restoration, climate change adaptation, and biodiversity conservation programs.
As global populations increase and land resources diminish, sustainable management of current natural resources becomes increasingly paramount. In order to design and implement better management practices, land resource managers need access to better and more site specific datasets on soils, vegetation, topography and climate. Collecting or accessing these datasets can be labor intensive, expensive, or even both. LandPKS is designed to help offset some of these challenges by not only making it easier for land managers to collect site specific soil and vegetation data, but also allowing them to access several global databases on soils, climate and topography. The combination of geo-referenced site characterization data and these global datasets will also eventually allow the system to provide estimates of potential productivity and risk of erosion, offering assessments of sustainability under various land use scenarios. To achieve this interface, the global LandPKS is being developed as an open-source suite of mobile phone apps connected to cloud-based global databases and models (Herrick et al. 2013; Herrick et al. 2016).
This system is designed to provide tools for:

• 1) collection, storage, access, and sharing of local and scientific data, information, and knowledge
• 2) selection and interpretation of management- and policy-relevant information to support decision making

The initial components of LandPKS are two integrated mobile apps for collecting

• (1) the basic soil and topographic information necessary to determine land potential (LandInfo)
• (2) soil and vegetation cover data necessary to assess and monitor major changes in plant community composition and wind and water erosion risk (LandCover).
• Further mobile apps are being developed to help collect information on Biomass (Biomass Monitor), Livestock body condition (Body Condition Score) as well as crops (Crop Monitor)

More detail information about this project.

Main web site : http://landpotential.org/

Portal Site: https://portal.landpotential.org/

Mobile app: Apple App Store    Google Play Store

Github repo :