Fine-Grained Arabic Post (Tweet) Geolocation Prediction Using Deep Learning Techniques
Leveraging Twitter data for crisis management necessitates the accurate, fine-grained geolocation of tweets, which unfortunately is often lacking, with only 1–3% of tweets being geolocated. This work addresses the understudied problem of fine-grained geolocation prediction for Arabic tweets, focusin...
Saved in:
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-01-01
|
| Series: | Information |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2078-2489/16/1/65 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832588340480704512 |
|---|---|
| author | Marwa K. Elteir |
| author_facet | Marwa K. Elteir |
| author_sort | Marwa K. Elteir |
| collection | DOAJ |
| description | Leveraging Twitter data for crisis management necessitates the accurate, fine-grained geolocation of tweets, which unfortunately is often lacking, with only 1–3% of tweets being geolocated. This work addresses the understudied problem of fine-grained geolocation prediction for Arabic tweets, focusing on the Kingdom of Saudi Arabia. The goal is to accurately assign tweets to one of thirteen provinces. Existing approaches for Arabic geolocation are limited in accuracy and often rely on basic machine learning techniques. Additionally, advancements in tweet geolocation for other languages often rely on distinct datasets, hindering direct comparisons and assessments of their relative performance on Arabic datasets. To bridge this gap, we investigate eight advanced deep learning techniques, including two Arabic pretrained language models (PLMs) on one constructed dataset. Through a comprehensive analysis, we assess the strengths and weaknesses of each technique for fine-grained Arabic tweet geolocation. Despite the success of PLMs in various tasks, our results demonstrate that a combination of Convolution Neural Network (CNN) and Long Short-Term Memory (LSTM) layers yields the best performance, achieving a test accuracy of 93.85%. |
| format | Article |
| id | doaj-art-4b3b85d25b094e839b32702b505c158d |
| institution | Kabale University |
| issn | 2078-2489 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Information |
| spelling | doaj-art-4b3b85d25b094e839b32702b505c158d2025-01-24T13:35:20ZengMDPI AGInformation2078-24892025-01-011616510.3390/info16010065Fine-Grained Arabic Post (Tweet) Geolocation Prediction Using Deep Learning TechniquesMarwa K. Elteir0Informatics Research Institute (IRI), City of Scientific Research and Technological Applications (SRTA-City), Alexandria 5220211, EgyptLeveraging Twitter data for crisis management necessitates the accurate, fine-grained geolocation of tweets, which unfortunately is often lacking, with only 1–3% of tweets being geolocated. This work addresses the understudied problem of fine-grained geolocation prediction for Arabic tweets, focusing on the Kingdom of Saudi Arabia. The goal is to accurately assign tweets to one of thirteen provinces. Existing approaches for Arabic geolocation are limited in accuracy and often rely on basic machine learning techniques. Additionally, advancements in tweet geolocation for other languages often rely on distinct datasets, hindering direct comparisons and assessments of their relative performance on Arabic datasets. To bridge this gap, we investigate eight advanced deep learning techniques, including two Arabic pretrained language models (PLMs) on one constructed dataset. Through a comprehensive analysis, we assess the strengths and weaknesses of each technique for fine-grained Arabic tweet geolocation. Despite the success of PLMs in various tasks, our results demonstrate that a combination of Convolution Neural Network (CNN) and Long Short-Term Memory (LSTM) layers yields the best performance, achieving a test accuracy of 93.85%.https://www.mdpi.com/2078-2489/16/1/65TwittergeolocationArabicdeep learningCNNLSTM |
| spellingShingle | Marwa K. Elteir Fine-Grained Arabic Post (Tweet) Geolocation Prediction Using Deep Learning Techniques Information geolocation Arabic deep learning CNN LSTM |
| title | Fine-Grained Arabic Post (Tweet) Geolocation Prediction Using Deep Learning Techniques |
| title_full | Fine-Grained Arabic Post (Tweet) Geolocation Prediction Using Deep Learning Techniques |
| title_fullStr | Fine-Grained Arabic Post (Tweet) Geolocation Prediction Using Deep Learning Techniques |
| title_full_unstemmed | Fine-Grained Arabic Post (Tweet) Geolocation Prediction Using Deep Learning Techniques |
| title_short | Fine-Grained Arabic Post (Tweet) Geolocation Prediction Using Deep Learning Techniques |
| title_sort | fine grained arabic post tweet geolocation prediction using deep learning techniques |
| topic | Twitter geolocation Arabic deep learning CNN LSTM |
| url | https://www.mdpi.com/2078-2489/16/1/65 |
| work_keys_str_mv | AT marwakelteir finegrainedarabicposttweetgeolocationpredictionusingdeeplearningtechniques |