Overview: Amr's recent research focuses on scalability and expressiveness for spatial data science. He particularly emphasizes enabling social scientists to analyze large-scale spatial data using advanced spatial statistical analysis techniques. Traditional computational systems for spatial data have primarily addressed basic queries like spatial kNN, spatial range, and spatial join queries, including variations such as kNN-join queries. However, these queries fall short of addressing the comprehensive needs of spatial statistical analysis. Statistical methods are extensively utilized by social scientists, including geographers, environmental scientists, and ecologists. Consequently, over the past two decades, there has been a parallel development of computing tools, resulting in Python and R libraries like PySAL, GeoDa, GeoPy, and GeoPandas, catering to this broad user base. These users are well-positioned to conduct meaningful analyses on the plethora of available datasets. Notably, ~14 million datasets indexed by Google Dataset Search, predominantly originating from geosciences and social sciences, constitute 45.2% of all datasets, accessed by a third of all users and primarily dominated by spatial data. However, current tools encounter scalability issues with substantial spatial data, such as the vast amounts of social media data rich in insights for social scientists. Amr’s work seeks to fill this void by empowering social scientists with scalable computing systems for spatial statistical analysis on large datasets. Additionally, it enhances the expressiveness of queries, enabling previously unfeasible queries due to scalability constraints.

Amr's recent work spans various aspects of scalable and expressive spatial data science. He has significantly enhanced the scalability and expressiveness of spatial grouping queries, which are foundational for spatial statistical analyses. His improvements cover the grouping of spatial points, lines, and polygons across diverse settings and applications. Notably, he has developed advanced spatial regionalization queries, essentially sophisticated polygon grouping queries, that surpass existing methods with up to a 200x increase in runtime speed and the capacity to handle datasets up to 10 times larger. Additionally, he has facilitated the generation of random distributions, empowering the statistical assessment of spatial regionalization solution quality. Side by side, Amr's work integrates machine learning to augment both the scalability and accuracy of spatial queries. This integration encompasses a variety of learning techniques to refine spatio-textual selectivity estimation in dynamic streaming data, identify spatial hotspots along networks with statistical rigor, and enhance the efficiency of spatial joins in distributed settings.

Interdisciplinary collaborations: Interdisciplinary projects, developed in partnership with social scientists, are at the heart of Amr's research, and pivotal to the mission of the UCR Center for Geospatial Sciences, which thrives as an interdisciplinary cluster of faculty members. His interdisciplinary projects include collaborators from University of California, Riverside (UCR), as well as the American Association of Geographers, Texas State University, San Diego State University, San Diego Mesa College, the California Geographic Alliance, and Sweetwater Union High School District.

Notable achievements: Amr's research is published and recognized in top research venues for big data management and spatial data analysis, including ACM SIGSPATIAL, IEEE ICDE, VLDB, SSTD, IEEE MDM, VLDB Journal, ACM TSAS, IEEE TKDE, Springer GeoInformatica, in addition to interdisciplinary venues. His papers are shortlisted for best paper awards at five major conferences for spatial and spatio-temporal data analysis, IEEE ICDE'14, twice at ACM SIGSPATIAL'19 and '23, IEEE MDM'23 and SSTD'23. He is also awarded the best paper at the 6th International Conference on Information and Communication Technologies for Disaster Management (ICT-DM'19). His research is being supported by eight research funding grants from the USA National Science Foundation (NSF), Google, and Microsoft, four as a Principal Investigator (PI), two as a co-PI and two as a collaborator. This includes the NSF CAREER Award 2023, a distinction that ranks among the most prestigious recognitions for assistant professors from NSF. So far, we raised total of 4.67M USD along with collaborators, including 1.24M USD dedicated to supporting his research group.

Scalable Spatial Data Science for Social Scientists

Through an interdisciplinary team of geographic information scientists and computer scientists, this project provides the next generation of scalable and high performance spatial analytical techniques. The main project goal is bridging two worlds: the spatial data management and geospatial analysis libraries that perform complex statistical analysis on geospatial datasets. This requires addressing the existing challenges that have hindered intensive utilization of geospatial data from different sources for more extensive analysis. To achieve this, the project will innovate new computational algorithms and combine them with scalable spatial and spatio-temporal data management techniques for regionalization, polygon-based queries, spatial-aware clustering, computational inference, and more.

CSforAll: Computational Thinking in Education

This work is based on a researcher-practitioner partnership (RPP) to articulate K-12 and college pathways that will expand opportunities for all students to develop geo-computational thinking skills. This pilot RPP is composed of geographers, computer science educators, and geospatial technology specialists experienced in serving underrepresented minority students and communities. This exploratory research is to inform educational standards and tested approaches to help institutions understand the capacity they need to modernize geography education. The efforts is being extended to other scientific curriculum, such as biology and chemistry.

Scalable Analysis on User-generated Data

This project innovates scalable techniques and systems for Microblogs data, e.g., tweets, online reviews on TripAdvisor, Yelp, or Amazon, user comments on news websites, or social media micro-posts like comments, check-ins and interactions. We produced system prototypes, Kite and Taghreed.The ultimate goal of these systems is to serve as a scalable backend that facilitate building applications on top of Microblogs data hiding all the data management details from the developers, just like database systems work as backend for relational data applications. Taghreed is already powering a startup company serving social media analysis services for Middle Eastern customers, initially incubated by Wadi Makkah innovation incubator.