1 Christoph F. Eick UH-DAIS Research Projects 8/19-8/20 - PowerPoint Presentation

Slide1

1

Christoph F.

Eick

UH-DAIS Research Projects 8/19-8/20

2

Spatial and

Spatio

-temporal Data Analysis Frameworks

St

^2—Tools for

Spatio-temporal Data StorytellingUnderstanding US Emotions in Time and SpaceIntelligent Crowdsourcing Collocation Mining FrameworksDisaster InformaticsMRI Image Analysis and Medical Informatics (lead by Dr. Tsekos)Scalable Algorithms for Interestingness Hotspot DiscoveryEducational Data Mining (lead by Dr. Rizk) Using AI for Route Planning Flood Forecasting and Flood Risk Assessment

Christoph F. Eick

Comment:

Active Projects

are in

white

,

yellow

and

blue.

Intelligent Crowdsourcing

Motivation

:

Crowdsourcing has become quite popular:

Companies use it to create datasets for machine learning and to outsource tasks to the

crowd

.OpenStreetMap and Waze are examples of crowdsourcing success stories.

Research Goals: Investigate AI Techniques to Enhance Crowdsourcing Develop Intelligent Crowdsourcing Apps for Disaster Informatics Investigate the psychological and social aspects of intelligent crowdsourcing

Section 4:



see other Slide Show

Data

Analysis and Intelligent Systems Lab

Great Dismal Swamp, Virginia

Tweet Emotion Mapping:Understanding US Emotions in Time and Space Related to: http://worldhappiness.report/ed/2018/ & http://hedonometer.org Inspired by: https://www.ted.com/talks/hans_rosling_asia_s_rise_how_and_whenChristoph F. Eick

UH-DAIS

K2

: Happiness and Opinion Mapping (from Tweets & …)

5

Amazing company values wow!

Happy Monday! Good start

Everyday in every way my life gets better & better

I

wanna

scream, I

wanna

shout as I’m not okay

Why do I hate myself so much

You make me sick

We lost ourselves

Another lovely day

I love you

I hate you

❤️

Emotion scores

Emotion scores

K2

Project Goals

Given a set of tweets (or questionnaires soliciting preferences and opinions) with the location (longitude and latitude), time they were posted and their emotional assessment in [-1,+1] (+1:=very positive emotions, 0:=no or even mix of emotions, -1: very negative emotions)

Research Steps and Goals

:

Subdivide the dataset into batches, corresponding to different time intervals

Identify spatial clusters of highly positive emotions (e.g. average emotional assessment >0.4) and regions of highly negative emotions (e.g. average emotion assessment <

-0.4) for each batch. Capture patterns of change and evolution of the regions identified in 2. Based on a selected story type and user preferences, convert results found in steps 2 and 3 into a narrative and animations that tells the story of spatio-temporal evolution of emotions in an region (e.g. Texas, US,…) over a period of time (e.g. 5 years, 1 year, 1 month), similar to: https://www.ted.com/talks/hans_rosling_asia_s_rise_how_and_when

Disaster Informatics

Active and Future Research Themes:

Intelligent Crowdsourcing Approaches for Disaster Help

Using Social Media Data During Disasters: Develop social media analysis tools that can be leveraged through crisis informatics and actionable policy steps nonprofits and government entities can take to integrate them into disaster response and recovery.

Develop Disaster Tweet Summarization and Change Analysis Frameworks

Collocation Mining Frameworks

Definition:

Spatial colocation patterns represent subsets of spatial events whose instances are often located in close geographic proximity. For example, car break-ins might often occur in close proximity of shopping malls.

Research Themes:

Density-based Collocation Mining Approaches

Spatio

-temporal Collocation Mining

Algorithms to Discover Regional Collocation Patterns

Intelligent Data Storytelling Tools

Motivation:

Communicating the story behind the data is a major challenge in most Data Science projects. Consequently, recently data storytelling has gained a lot of attention in the Commercial Data Science Community. Data storytelling is a structured approach for communicating data insights and combines three key elements: data, visuals, and narrative.

Objectives:

This research centers on development of automated, intelligent data storytelling tools; in particular, it centers on the design and implementation of a

spatio

-temporal data storytelling framework called

Kilimanjaro. Happiness Mapping is used as a case study in this project to demo the frameworks and tools we develop.

Fast

Interestingness Hotspot Discovery

Objective

:

Find interesting contiguous regions in spatial data sets based on the domain expert’s notion of interestingness which is captured in an interestingness function

Methodology

: Transform Dataset Into GraphsIdentify hotspot seedsGrow seeds by adding neighboring objectsRemove redundant hotspots using a graph-based approachFind Scope of hotspots (polygonal boundary detection)Data sets: Gridded, polygonal, point-based data sets

Educational Data Mining (EDM)

UH-DMML

Polygon Analysis for Better Flood Risk Mapping

Christoph F.

Eick

Austin Fire

First Response

Vehicle Flood

Risk Map

Hand Value Multi-ContourMapsFEMA Flood Risk ZonesFind CorrespondenceFind Agreement,Combine, Validate,EvaluateDEM (Digital Elevation Maps)

HEC-RAS

Generated

Polygons

Knowledge of Flooded Areas

from Past Floods

UH-DAIS

Helping Scientists to Make Sense Out of their Data

Figure 1: Co-location regions involving deep and

shallow ice on Mars

Figure 2: Interestingness hotspots where both income and CTR are high.

Figure 3: Maryland Crime Hotspots

UH-DAIS

Recent Contributors to UH-DAIS Research

UH-DAIS

PhD Students

:

Yongli

Zhang,

Romita

Banerjee, Chong Wang and Karima Elgarroussi. Master Students: Yue Cao, Anusha Nemilidinne, Anjana Kumari, Priyal Kulkarni, Qian Qiu, Arjun SV and Akhil Talari. Undergraduate Students: Deniz Burduroglu, Duong Nguyen, Victor Zeng, Yilei Tian, Pallovi Romero, Israel Perez, Jackson MurrellVisiting/Exchange Students: Khadija Khaldi Contributing Alumni: Sujing Wang and Paul Amalaman.

Some UH-DAIS Graduates 1

Christoph F.

Eick

Dr. Wei Ding, Associate Professor, Department of Computer Science,

University of Massachusetts

, Boston

Sharon M. Tuttle, Professor,

Department of Computer Science,Humboldt State University, Arcata, California

Christopher T.

Ryu

, Professor,

Department of Computer Science,

California State University

, Fullerton

Sujing

Wang, Assistant Professor,

Department of Computer Science,

Lamar University

, Beaumont, Texas

Some UH-DAIS Graduates 2

Christoph F. Eick

Yongli

Zhang

PhD Airbnb

Chun-sheng Chen

PhD eBay

Puja Anchlia MS eBayChong Wang MS AppleJustin Thomas MS John Hopkins University Applied Physics LaboratoryMei-kang Wu MS Microsoft, Bellevue, Washington Jing Wang MS AOL, CaliforniaRachsuda Jiamthapthaksin PhD Faculty, Assumption University, Bangkok, Thailand

Data Driven Flood Forecasting Frameworks

17