Applicability Analysis and Ensemble Application of BERT with TF-IDF, TextRank, MMR, and LDA for Topic Classification Based on Flood-Related VGI DOI Creative Commons
Wenying Du, Chang Ge, Shuang Yao

et al.

ISPRS International Journal of Geo-Information, Journal Year: 2023, Volume and Issue: 12(6), P. 240 - 240

Published: June 9, 2023

Volunteered geographic information (VGI) plays an increasingly crucial role in flash floods. However, topic classification and spatiotemporal analysis are complicated by the various expressions lengths of social media textual data. This paper conducted applicability on bidirectional encoder representation from transformers (BERT) four traditional methods, TextRank, term frequency–inverse document frequency (TF-IDF), maximal marginal relevance (MMR), linear discriminant (LDA), results show that for user type, BERT performs best Government Affairs Microblog, whereas LDA-BERT We Media Microblog. As text length, TF-IDF-BERT works better texts with a length <70 >140 words, 70–140 words. For evolution pattern, study suggests Henan rainstorm, topics follow general pattern “situation-tips-rescue”. Moreover, this detected hotspot “Metro Line 5” related to rainstorm discovered topical focus spatially shifts Zhengzhou, first Xinxiang, then Hebi, showing remarkable tendency south north, which was same as report issued authorities. integrated multi-methods improve overall accuracy Sina microblogs, facilitating flooding.

Language: Английский

Urban traffic flow prediction: a dynamic temporal graph network considering missing values DOI
Peixiao Wang, Yan Zhang, Tao Hu

et al.

International Journal of Geographical Information Science, Journal Year: 2022, Volume and Issue: 37(4), P. 885 - 912

Published: Nov. 17, 2022

Accurate traffic flow prediction on the urban road network is an indispensable function of Intelligent Transportation Systems (ITS), which great significance for planning. However, current methods still face many challenges, such as missing values and dynamic spatial relationships in flow. In this study, a temporal graph neural considering (D-TGNM) proposed prediction. First, inspired by Bidirectional Encoder Representations from Transformers (BERT), we extend classic BERT model, called Traffic BERT, to learn associations structure. Second, propose (TGNM) mine patterns data scenarios Finally, D-TGNM model can be obtained integrating learned into TGNM model. To train design novel loss function, considers problem flow, optimize The was validated actual dataset collected Wuhan, China. Experimental results showed that achieved good under four (15% random missing, 15% block 30% missing), outperformed ten existing state-of-the-art baselines.

Language: Английский

Citations

29
A systematic review of natural language processing applications for hydrometeorological hazards assessment DOI Open Access
Achraf Tounsi, Marouane Temimi

Natural Hazards, Journal Year: 2023, Volume and Issue: 116(3), P. 2819 - 2870

Published: Feb. 8, 2023

Language: Английский

Citations

22
Dynamic risk assessment of urban flood disasters based on functional area division—A case study in Shenzhen, China DOI
Ting Wang, Huimin Wang, Zhiqiang Wang

et al.

Journal of Environmental Management, Journal Year: 2023, Volume and Issue: 345, P. 118787 - 118787

Published: Aug. 26, 2023

Language: Английский

Citations

22
An approach of using social media data to detect the real time spatio-temporal variations of urban waterlogging DOI
Yilin Chen, Maochuan Hu, Xiaohong Chen

et al.

Journal of Hydrology, Journal Year: 2023, Volume and Issue: 625, P. 130128 - 130128

Published: Sept. 10, 2023

Language: Английский

Citations

18
Attribution analysis of urban social resilience differences under rainstorm disaster impact: Insights from interpretable spatial machine learning framework DOI

Tianshun Gu,

Hongbo Zhao, Yue Li

et al.

Sustainable Cities and Society, Journal Year: 2024, Volume and Issue: unknown, P. 106029 - 106029

Published: Dec. 1, 2024

Language: Английский

Citations

8
Toponym resolution leveraging lightweight and open-source large language models and geo-knowledge DOI Creative Commons
Xuke Hu, Jens Kersten, Friederike Klan

et al.

International Journal of Geographical Information Science, Journal Year: 2024, Volume and Issue: unknown, P. 1 - 28

Published: Sept. 24, 2024

Language: Английский

Citations

7
Revealing urban vibrancy stability based on human activity time-series DOI
Jiani Ouyang, Hong Fan, Luyao Wang

et al.

Sustainable Cities and Society, Journal Year: 2022, Volume and Issue: 85, P. 104053 - 104053

Published: July 10, 2022

Language: Английский

Citations

23
Migratable urban street scene sensing method based on vision language pre-trained model DOI Creative Commons
Yan Zhang, Fan Zhang, Nengcheng Chen

et al.

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2022, Volume and Issue: 113, P. 102989 - 102989

Published: Sept. 1, 2022

We propose a geographically reproducible approach to urban scene sensing based on large-scale pre-trained models. With the rise of GeoAI research, many high-quality observation datasets and deep learning models have emerged. However, geospatial heterogeneity makes these resources challenging share migrate new application scenarios. This paper introduces vision language semantic model for street view image analysis as an example. bridges boundaries data formats under location coupling, allowing acquisition text-image objective descriptions in physical space from human perspective, including entities, entity attributes, relationships between entities. Besides, we proposed SFT-BERT extract text feature sets 10 land use categories 8,923 scenes Wuhan. The results show that our method outperforms seven baseline models, computer vision, improves 15% compared traditional methods, demonstrating potential pre-train & fine-tune paradigm GIS spatial analysis. Our could also be reused other cities, more accurate judgments obtained by inputting images different angles. code is shared at: github.com/yemanzhongting/CityCaption.

Language: Английский

Citations

23
How can voting mechanisms improve the robustness and generalizability of toponym disambiguation? DOI Creative Commons
Xuke Hu, Yeran Sun, Jens Kersten

et al.

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2023, Volume and Issue: 117, P. 103191 - 103191

Published: Feb. 1, 2023

Natural language texts, such as tweets and news, contain a vast amount of geospatial information, which can be extracted by first recognizing toponyms in texts (toponym recognition) then identifying their representations disambiguation). This paper focuses on toponym disambiguation, approached resolution entity linking. Recently, many novel approaches, especially deep learning-based, have been proposed, CamCoder, GENRE, BLINK. However, these approaches were not compared the same large datasets. Moreover, there is still need space to improve robustness generalizability further. To mitigate two research gaps, this paper, we propose spatial clustering-based voting approach combining several individual compare ensemble with 20 latest commonly-used based 12 public datasets, including highly challenging datasets (e.g., WikToR). They are six types: tweets, historical documents, web pages, scientific articles, Wikipedia containing 98,300 toponyms. Experimental results show that performs best all achieving an average [email protected] 0.86, proving its robustness. It also drastically improves performance resolving fine-grained places, i.e., POIs, natural features, traffic ways. The detailed evaluation inform future methodological developments guide selection proper application needs.

Language: Английский

Citations

14
Appraisal of Urban Waterlogging and Extent Damage Situation after the Devastating Flood DOI
Shan‐e‐hyder Soomro, Muhammad Waseem Boota, Xiaotao Shi

et al.

Water Resources Management, Journal Year: 2024, Volume and Issue: 38(12), P. 4911 - 4931

Published: June 8, 2024

Language: Английский

Citations

5