2022
DOI: 10.1111/tgis.12902
|View full text |Cite
|
Sign up to set email alerts
|

ChineseTR: A weakly supervised toponym recognition architecture based on automatic training data generator and deep neural network

Abstract: Toponym recognition is used to extract toponyms from natural language texts, which is a fundamental task of ubiquitous geographic information applications. Existing toponym recognition methods with state‐of‐the‐art performance mainly leverage supervised learning (i.e., deep‐learning‐based approaches) with parameters learned from massive, labeled datasets that must be annotated manually. This is a great inconvenience when model training needs to fit different domain texts, especially those of social media messa… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
7
0

Year Published

2022
2022
2024
2024

Publication Types

Select...
9

Relationship

2
7

Authors

Journals

citations
Cited by 18 publications
(7 citation statements)
references
References 49 publications
(82 reference statements)
0
7
0
Order By: Relevance
“…To address this, Ma et al proposed a BERT-BiLSTM-CRF deep neural network architecture for Chinese text geotagging tasks [33]. Qiu et al proposed a ChineseTR architecture based on weakly supervised BERT + BiLSTM + CRF for Chinese geotagging and trained the Chinese geotagging model on a training dataset generated from the People's Daily corpus [34]. Tao et al proposed an improved BERT model method for geographical named entity recognition and verified the effectiveness of the method [35].…”
Section: Methods Based On Deep Neural Networkmentioning
confidence: 99%
“…To address this, Ma et al proposed a BERT-BiLSTM-CRF deep neural network architecture for Chinese text geotagging tasks [33]. Qiu et al proposed a ChineseTR architecture based on weakly supervised BERT + BiLSTM + CRF for Chinese geotagging and trained the Chinese geotagging model on a training dataset generated from the People's Daily corpus [34]. Tao et al proposed an improved BERT model method for geographical named entity recognition and verified the effectiveness of the method [35].…”
Section: Methods Based On Deep Neural Networkmentioning
confidence: 99%
“…Place name recognition is vital to the georeferencing and enrichment of geospatial gazetteers. However, it is a time-consuming task and needs massive labeled datasets [26]. In most cases, there is even no place name within the text.…”
Section: Related Workmentioning
confidence: 99%
“…Of course, during sentence formation, two important problems arise: determining which words or phrases should be included in the sentence and determining how those words or phrases are arranged in the sentence. To address these issues, we adopt and comply with unigram language model theory in our language model, in which the sequence of words or phrases in a sentence is independent (Qiu, Xie, Wang, et al, 2022; Tripathy et al, 2016). A standard statistical language model aims to calculate the probability p ( w 1 , w 2 , … w n ) over a sequence of words in a sentence.…”
Section: Computational Frameworkmentioning
confidence: 99%
“…The exponential growth of natural language text data, such as social media messages, has supported rich sources of geographic information. Many types of spatial relations are mentioned in these data, which are frequently described by natural language spatial relation (NLSR) terms or predicative phrases/words to represent and share relational knowledge and spatial locations about spatial objects (Du et al, 2015, 2016; Du & Guo, 2016; Qiu, Xie, Wang, et al, 2022; Wu et al, 2022). Therefore, research has focused on the NLSR description, which contains not only significant user‐friendly searches and interactions, but also various retrieved geographic information (Hu et al, 2017; Ma, 2022; Smole et al, 2011; Zheng et al, 2022) and disaster management information (Du et al, 2018).…”
Section: Introductionmentioning
confidence: 99%