Mabruri, Iqbal (2019) Sistem Deteksi Kecelakaan Melalui Pemanfaatan Sosial Media Twitter Berbasis Deep Learning. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Text
05111540000120-Iqbal Mabruri-Buku_TA.pdf - Accepted Version Restricted to Repository staff only until 1 October 2023. Download (2MB) | Request a copy |
Abstract
Menurut WHO kecelakaan jalan raya merupakan salah satu dari 10 penyebab kematian tebesar di dunia. Banyak dari korban meninggal akibat dari penangan yang terlambat. Oleh sebab itu perlu adanya sistem yang dapat mendeteksi dini kecelakaan, agar dapat mempercepat penanganan. Banyak riset sebelumnya yang memanfaatkan tweet dari Twitter untuk mendeteksi kecelakaan. Namun deteksi kecelakaan saja tidak cukup perlu adanya informasi lokasi kecelakaan dan memastikan setiap kecelakaan yang dilaporkan bersifat unik. Pada tugas akhir ini dibuatlah sebuah sistem yang dapat mendeteksi kecelakaan, mengetahui lokasi, dan mencegah adanya laporan kecelakaan duplikat. Untuk dapat mendeteksi kecelakaan digunakan metode deep learning arsitektur gated recurrent unit (GRU). Di artikel ini diusulkan fungsi filter yang dapat mencegah adanya deteksi kecelakaan duplikat. Dalam mengembangkan sistem, baik tweet kecelakaan maupun bukan dikumpulkan dari Juni 2018 hingga Agustus 2018. Dari hasil uji coba yang dilakukan model deep learning memberikan hasil akurasi 0.974 dan f measure 0.872 pada data validasi, akan tetapi model ini mengalami gejala overfitting ketika diuji menggunakan data real time.
================================================================================================
Based on WHO’s data injury from traffic accident is one of the ten top cause of death in the world. Many of these death is because of late response. Therefore, a system that can detect early traffic acci-dentisneeded. Many of the previous research use tweet from Twitter as the data source to detect traffic accident. However, only able to detect traffic accident is not enough. Giving additional informa-
tion, such as, location is important, and make sure each accident detected is unique. In this undergraduate thesis a system that can detect accident from tweet,giving the accident location, and prevent duplicate accident detection is developed. The detection method the system use in this undergraduate thesis is using deep learning ar-
chitecturegatedre current unit(GRU).During the system’sdevelop-ment first tweet are collected from June 2018 toAugust 2018. Later the collected tweet is used to train a deep learning model with ar-chitecture of Gated Recurrent Unit (GRU) using word embedding as feature. Although the model manages to get 0.974 accuracy and 0.872 f measure, the model shows overfitting symptom.
Item Type: | Thesis (Undergraduate) |
---|---|
Additional Information: | RSIf 006.754 1 Mab s-1 2019 |
Uncontrolled Keywords: | weet, Kecelakaan, GRU, Word Embedding, Traffic Accident |
Subjects: | H Social Sciences > HE Transportation and Communications > HE5614.3.N5 Traffic accidents T Technology > T Technology (General) > T57.5 Data Processing |
Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis |
Depositing User: | Iqbal Mabruri |
Date Deposited: | 23 Sep 2021 09:31 |
Last Modified: | 23 Sep 2021 09:31 |
URI: | http://repository.its.ac.id/id/eprint/60694 |
Actions (login required)
View Item |