Artificial Intelligence and Deep Learning Symposium: Geospatial Semantics and Geo-Text Data Analytics

Type: Paper
Theme:
Sponsor Groups: Geographic Information Science and Systems Specialty Group, Spatial Analysis and Modeling Specialty Group, Cyberinfrastructure Specialty Group
Poster #:
Day: 4/12/2018
Start / End Time: 1:20 PM / 3:00 PM
Room: Grand Ballroom A, Astor, 2nd Floor
Organizers: Yingjie Hu, Kevin Sparks
Chairs: Yingjie Hu

Description

Geospatial semantics is a broad field that involves research ranging from geo-ontology engineering and gazetteer construction to geographic information retrieval (GIR) and place semantics. In recent years, there is an increasing number of datasets that contain both geographic locations and natural language texts, such as geotagged tweets, geotagged Wikipedia pages, geotagged housing advertisements, as well as daily descriptions and news articles that contain place names. These geo-text data offer new opportunities for research, since there is a variety of information that can be extracted from texts, such as place names, various entities, topics, opinions, and emotions, and we can examine the spatiotemporal patterns of the extracted information. Accordingly, mining geo-text data often needs an effective integration of geospatial analysis and natural language processing (NLP). With the recent advancements of AI and deep learning, new methods, such as Recurrent Neural Network (RNN), have also been employed to analyze texts and have received promising results. A combination of geo-text data with deep learning and other AI techniques may give birth to even more ideas. Applying these methods in fields such as disaster response, land use mapping, public health, and population distribution (etc.) can provide novel and impactful solutions to existing and new geographic problems. Fundamentally, geo-text data analytics is situated under the umbrella of geospatial semantics, since effective data analysis needs an understanding of the semantics of texts (e.g., the topics, mentioned entities, and emotions) and its association with geospatial locations. This special session aims to bring together researchers with similar interests to exchange ideas and discuss future directions.

Example topics (include but not limited to):
- Spatial, temporal, and semantic analytics on geotagged social media data
- Deep learning or other techniques for extracting and disambiguating place names from texts
- Deep learning or other techniques for extracting spatial relations from texts
- Deep learning or other techniques for geo-text data classification
- Sentiment analysis and geospatial clustering based on geo-text data
- Topic modeling and geospatial analysis with geo-text data
- Gazetteer construction with place entries extracted from geo-text data
- New approaches for integrating geospatial analysis with natural language processing
- …


To present your research in our session, please submit your abstract through AAG website (http://www.aag.org/cs/annualmeeting/register) and send your PIN to Yingjie Hu (yhu21@utk.edu) or Kevin Sparks (sparkska@ornl.gov) by October 25, 2017.



Agenda

Type Details Minutes Start Time
Presenter Tuomo Hiippala*, Digital Geography Lab, University of Helsinki, Christoph Fink, Digital Geography Lab, University of Helsinki, Vuokko Heikinheimo, Digital Geography Lab, University of Helsinki, Henrikki Tenkanen, Digital Geography Lab, University of Helsinki, Tuuli Toivonen, Digital Geography Lab, University of Helsinki, Applying deep learning to multimodal data in social media 20 1:20 PM
Presenter Cheng Fu*, University of Maryland - College Park, Grant McKenzie, University of Maryland - College Park, Vanessa Frias-Martinez, University of Maryland - College Park, Kathleen Stewart, University of Maryland - College Park, Identifying spatiotemporal urban activities through linguistic signatures 20 1:40 PM
Presenter Bo Yan*, University of California, Santa Barbara, Krzysztof Janowicz, University of California, Santa Barbara, Gengchen Mai, University of California, Santa Barbara, Song Gao, University of Wisconsin, Madison, From ITDL to Place2Vec - Reasoning About Place Type Similarity and Relatedness by Learning Embeddings From Augmented Spatial Contexts 20 2:00 PM
Presenter Shaohua Wang*, UCSB, Yang Zhong, Claremont Graduate University , Hao Lu, SuperMap , Hui Guo, Institute of Geographic Sciences and Natural Resources Research, CAS, Zheng Long, SuperMap, Liang Long, Institute of Geographic Sciences and Natural Resources Research, CAS, A Chinese Address Geocoding Method Based on Automata Word Segmentation 20 2:20 PM
Presenter Morteza Karimzadeh*, Ohio State University, Alan M. MacEachren, Pennsylvania State University, Scott Pezonowski, Pennsylvania State University, Evaluating GeoTxt, a Scalable Geoparser for Toponym Recognition and Resolution 20 2:40 PM

To access contact information login