eScience through the Integration of Data and Models: A Biodiversity - - PowerPoint PPT Presentation

eScience through the Integration of Data and Models: A Biodiversity Scenario

Experience Paper Natalia Villanueva-Rosales *, Luis Garnica Chavira *, Nicholas del Rio †, Deana Pennington *

*Cyber-ShARE Center of Excellence, University of Texas at El Paso, El Paso, US

† Air Force Research Lab, Information Directorate, Rome, US.

Overview

▪ Motivation ▪ The Earth Life and Semantic Web Project (ELSEWeb) ▪ Creating Species Distribution Models: A User’s Perspective ▪ ELSEWeb’s Service Oriented Architecture ▪ ELSEWeb’s User Interface ▪ User Study ▪ Lessons Learned ▪ Conclusions and Future Work ▪ Acknowledgements 2

Motivation

▪ Large amount of time invested in finding and preprocessing data to use specialized tools. ▪ Heterogeneity of data formats, programming, query languages, and lack of documentation, contribute to the challenges of interoperability between scientific tools. ▪ It is expected to have a wide array of tools and data available for the generation of biodiversity models, the challenge is to integrate them ▪ Cyberinfrastructure (CI) streamlining the integration of data and models would allow scientists to focus on the science.

3

The Earth Life and Semantic Web Project (ELSEWeb)

▪ Aims to develop and test generic scalable solutions to address integration of data and models. ▪ Initial use case: Integrating raster-based environmental data sets such as satellite imagery with species distribution models (SDMs). ▪ This presentation: ELSEWeb architecture (including GUI), usability study and lessons learned.

4

Species Distribution Models (SDMs)

5

Agave Lecheguilla Output Lifemapper Model EDAC Environmental Data Agave Lecheguilla Occurrence Sets

COMPARISON MATRIX

User task without ELSEWeb ELSEWeb task User task with ELSEWeb

Search for rainfall, terrain

• r vegetation data
• Download data
• Import to GIS or Image

Processing System

• Obtain Lifemapper data

requirements

• Transform data
• Export data
• Import data to QGIS
• Install Lifemapper Plugin
• Select Lifemapper

experiment elements

• Assemble package
• Run Lifemapper

experiments

• TASKS REQUIRED FOR THE GENERATION

OF SDMS WITH AND WITHOUT THE USE OF ELSEWEB

6

ELSEWeb’s Service Oriented Architecture

▪ Four principles of the Model Web: open access, minimal barriers to entry, service-driven and scalable. ▪ Service Oriented Architecture (SOA) using Web Services to achieve interoperability between third party service providers. ▪ Semantic capabilities.

7

ELSEWeb’s Service Oriented Architecture

SOA View - Primary Representation

Semantic-Based Services

Layered View – Primary Representation

ELSEWeb’s GUI

Layered View – Primary Representation

ELSEWeb’s JavaScript Framework

Layered View – Primary Representation

Step 1

Region Bounding Box Selection

Step 2

Species Occurrence Set Selection

Step 3

Environmental Dataset Filtering

Step 4

Modeling Algorithm Selection

Step 5

Experiment Submission and Dataset Selection

Step 6

Experiment Results and Provenance

User Study

▪ Guidelines and methodologies provided by the U.S. Department of Health & Human Services (usability.gov). ▪ Sample of 15 students from the University of Texas at El Paso. ▪ The session captured participant’s navigational choices, logger observations, task completion rates and post- test survey. ▪ Screen interaction and participant audio was recorded with BB Flash Back Express Recorder software. ▪ Each individual session lasted approximately 15 minutes.

18

User Study

Evaluation Tasks

ID ¡ Task ¡

T1 ¡ Login to the website. ¡ T2 ¡ Establish a boxed region for dataset availability. ¡ T3 ¡ Select species occurrence set. ¡ T4 ¡ Set filtering parameters for corresponding datasets ¡ T5 ¡ Select Model algorithm and parameter values. ¡ T6 ¡ Select datasets for experiment submission. ¡ T7 ¡ Submit experiment specification. ¡ T8 ¡ Consult experiment status. ¡ 19

Usability Study Results

Task # Task short description #of participants completing the task Non- critical errors Task execution

• avg. time

(seconds)

1 Website login. 15 3 53 2 Establish boxed region for data availability. 15 11 94 3 Select species

• ccurrence set.

15 7 47 4 Set filtering parameters for datasets 15 9 73 5 Select model algorithm and parameter values. 14 19 6 Select datasets. 15 21 7 Submit experiment specification. 15 37 8 Consult experiment status. 13 6 34

Task Summary Table

20

Survey Sample Results

2 4 6 8 10 Poor Fair Average Good Excellent

Overall GUI Score

Responses

2 4 6 8 10 12

Strongly Disagree Disagree Neutral Agree Strongly Agree

Ease of use

Responses

21

User Issues

Good… ▪ “Interface was very useful and friendly.” ▪ “Overall, the site worked perfectly fine, was straight forward, and easy to use.” Not so good… ▪ “It was frustrating not knowing that an option must be modified and reconfigured in order to prepare the Environmental Data Filters section.” ▪ “I had a hard time at the beginning to figure out where the tabs where, I had to scroll down to figure out there were at the bottom.”

22

Recommendations

Overall Experiment Interface. ¡ Visual queues for the workflow been executed. ¡

Region

1

Species

2

Datasets

3

Algorithm

4

Selection

5

Submit

6 What I’m I missing to move to the next step?

23

General Description Parameters and Meanings

Related Work

▪ Model Web – aims for seamless interoperability between data and model

• providers. ELSEWeb focuses on environmental data into SDMs. (S. Nativi, et

al., 2013) ▪ Research Objects (ROs), semantically rich aggregation of resources produced and consumed by services. RO features in ELSEWeb: aggregation, identity, metadata and lifecycle. (S. Bechhofer, et al., 2013) ▪ IPlant collaborative framework, real time services cover for construction and execution of scientific workflows. In addition, ELSEWeb offers service

• rchestration. These systems can achieve interoperability through ontology
• mapping. (D. D. Gessler , et al., 2013)

24

Discussion and Lessons Learned

▪ ELSEWeb architecture is extensible. ▪ Use of standards and best practices as a key element to enable interoperability with other systems. ▪ Notifications and synchronizations with third-party services is essential when reusing and linking resources. ▪ Use of conceptual models for communication across members of the collaborative, multidisciplinary team. ▪ Involvement of end-users facilitated development of framework but also validation of technically sound results.

25

Future Work

26

▪ Extending ELSEWeb to incorporate process models for water resource analysis – Model to model integration. ▪ URI naming conventions. ▪ Implementing the recommendations obtained in from the usability study. ▪ Further development of user-centered interfaces. ▪ Scientific evaluation.

Acknowledgements

Linking knowledge across disciplines, data and models

Cyber-ShARE

• ELSEWeb was funded by the NASA ACCESS grants

NNX12AF49A (UTEP), NNX12AF52A (UNM), and NNX12AF45A (KU).

• This work used resources from Cyber-ShARE Center
• f Excellence supported by NSF grant HRD-0734825.

27

Useful Links

▪ ELSEWeb GUI: http://elseweb.cybershare.utep.edu/experimentgui ▪ Usability Study Report: http://elseweb.cybershare.utep.edu/publications ▪ Previous Papers: http://elseweb.cybershare.utep.edu/publications ▪ Usability.gov: http://www.usability.gov/ 28

Looking for collaborators, contact us!

The Earth Life and Semantic Web Project

http://elseweb.cybershare.utep.edu/

29