Using Bidirectional Encoder Representations from Transformers (BERT) to classify traffic crash severity types

Oliaee, Amir Hossein; Das, Subasish; Liu, Jinli; Rahman, M. Ashifur

doi:10.1016/j.nlp.2023.100007

Cited by 20 publications

(6 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Recently, advanced language models have been adopted in several studies to solve transportation science problems ( 25 – 28 ). This research aims to use a small data set to obtain a good prediction for a crash type classification problem using an advanced language model.…”

Section: Discussionmentioning

confidence: 99%

Applying Few-Shot Learning in Classifying Pedestrian Crash Typing

Weng

Das

Paal

2023

Transportation Research Record: Journal of the Transportation R

Self Cite

View full text Add to dashboard Cite

Pedestrian deaths account for 23% of all road traffic fatalities worldwide. After declining for three decades, pedestrian fatalities in the United States have been increasing with 6,941 fatalities in 2020, the highest number for more than two decades, impeding progress toward a zero-deaths transportation system. The Pedestrian and Bicycle Crash Analysis Tool (PBCAT) was developed to describe the pre-crash actions of the parties involved to better define the sequence of events and precipitating actions that lead to crashes involving motor vehicles and pedestrians or cyclists. Undoubtedly, police crash data influence decision-making processes in the transportation agencies. Using crash data from three major cities in Texas (during the period from 2018 to 2020), this study assessed the data quality of police-reported crash narratives on pedestrian-involved traffic crashes. As the pedestrian crash typing involves many categories, conventional machine-learning algorithms will not be sufficient in solving the classification problem from narrative texts. This study used few-shot learning (FSL), an advanced machine learning, to solve this issue. Using the pre-knowledge obtained from five different crash types and a few labeled data points of three unseen new crash types, the proposed model achieved roughly 40% overall accuracy. Also, four different configurations of crash types were formed and tested which indicates that the model is robust.

show abstract

Section: Discussionmentioning

confidence: 99%

Applying Few-Shot Learning in Classifying Pedestrian Crash Typing

Weng

Das

Paal

2023

Transportation Research Record: Journal of the Transportation R

Self Cite

View full text Add to dashboard Cite

show abstract

“…Evaluasi metriks digunakan untuk mengukur akurasi, recall, presisi, dan f1-score dari hasil klasifikasi atau perbandingan metode yang digunakan, berikut persamaannya [23], [24]:…”

Section: Evaluasiunclassified

“…merupakan pengukuran yang dihasilkan dari hasil pemrosesan atau perhitungan presisi dan recall dengan mempertimbangkan kelas hasil prediksi dengan kelas sesungguhnya pada data menggunakan persamaan 8 dibawah ini. True Positive, TF adalah True False, FP adalah False Positive, FN adalah False Negative, 𝑙 merupakan total kelas pada data yang dilatih, sedangkan 𝑛 𝑖 merupakan total data yang dilatih dari setiap kelas[23],[24].P-ISSN : 2089-3353 Volume 13 No. 2 | Agustus 2023: 326-333 E-ISSN : 2808-9162 Author : Ardiansyah 1) , Adika Sri Widagdo 2) , Krisna Nuresa Qodri 3Collection Dataset Pada penelitian yang telah dilakukan proses pengumpulan dataset, peneliti mendapatkan data dari ulasan di Google Maps puskesmas dan rumah sakit swasta ataupun negri yang berada pada daerah X. Penggunaan selenium, beautifulsoup menjadi tools yang digunakan dalam pengumpulan data perlu dilakukan dikarenakan ulasan pada Google maps menggunakan lazyload.…”

unclassified

Analisis sentimen terhadap pelayanan Kesehatan berdasarkan ulasan Google Maps menggunakan BERT

Ardiansyah,

Adika Sri Widagdo,

Krisna Nuresa Qodri

et al. 2023

View full text Add to dashboard Cite

The utilization of technology has developed in various scientific fields, without exception in health. Hospitals, health centers, and clinics are part of the health sector. Thus, it must evolve according to health service standards and patient measures or service user satisfaction that needs to be measured using sentiment analysis. The Media to give opinions to Health service providers is Google Maps. However, the anomaly is that the reviews and the given text are sometimes not correlated. Thus, The utilization of sentiment analysis using the scientific branch of artificial intelligence, namely Natural Language Processing (NLP), is an effective way to infer opinions. The research concluded that the BERT indobenchmark/indobert-base-p1 model has good performance to use of Indonesian text classification with a dataset of 4228 data after preprocessing, which at the beginning of the collection process obtained data as much as 4748 data. Split datasets into 3 data, namely training, validation, and test data, with a ratio of 70:30:30. The experimental results, The researchers found that the model allows the use of the model with other Indonesian texts. The results are 0.85 for accuracy and weighted avg, and macro avg 0.75 on the validation data training process. While the testing data training process is 0.86 for accuracy and weighted avg, the macro avg 0.73. In addition, researchers found that services are the most frequent topic in Health Services. Even though health services have improved, positive sentiment is the highest compared to other sentiment classes.

show abstract

“…Using Bidirectional Encoder Representations from Transformers (BERT) to classify traffic crash severity types [278]. Contextual Embeddings based on Fine-tuned Urdu-BERT for Urdu threatening content and target identification [271].…”

Section: Fig 5 Ai Components For Nlpmentioning

confidence: 99%

“…These models hold great promise for enhancing text-based interactions and insights. [245], [180], [188], [251], [261], [264], [271], [272], [274], [276], [277], [278], [279], [218], [288], [296], [297], [236], [299], [238] GPT Variants…”

Section: Fig 5 Ai Components For Nlpmentioning

confidence: 99%

Evolving Conversations: A Review of Chatbots and Implications in Natural Language Processing for Cultural Heritage Ecosystems

Suryanto,

Wibawa,

Hariyono

et al. 2023

IJRCS

View full text Add to dashboard Cite

Chatbot technology, a rapidly growing field, uses Natural Language Processing (NLP) methodologies to create conversational AI bots. Contextual understanding is essential for chatbots to provide meaningful interactions. Still, to date chatbots often struggle to accurately interpret user input due to the complexity of natural language and diverse fields, hence the need for a Systematic Literature Review (SLR) to investigate the motivation behind the creation of chatbots, their development procedures and methods, notable achievements, challenges and emerging trends. Through the application of the PRISMA method, this paper contributes to revealing the rapid and dynamic progress in chatbot technology with NLP learning models, enabling sophisticated and human-like interactions on the trends observed in chatbots over the past decade. The results, from various fields such as healthcare, organization and business, virtual personalities, to education, do not rule out the possibility of being developed in other fields such as chatbots for cultural preservation while suggesting the need for supervision in the aspects of language comprehension bias and ethics of chatbot users. In the end, the insights gained from SLR have the potential to contribute significantly to the advancement of chatbots on NLP as a comprehensive field.

show abstract

Using Bidirectional Encoder Representations from Transformers (BERT) to classify traffic crash severity types

Cited by 20 publications

References 21 publications

Applying Few-Shot Learning in Classifying Pedestrian Crash Typing

Applying Few-Shot Learning in Classifying Pedestrian Crash Typing

Analisis sentimen terhadap pelayanan Kesehatan berdasarkan ulasan Google Maps menggunakan BERT

Evolving Conversations: A Review of Chatbots and Implications in Natural Language Processing for Cultural Heritage Ecosystems

Contact Info

Product

Resources

About