KLASIFIKASI JENIS PENGADUAN DAN VISUALISASI LOKASI DARI CRAWLING DATA TWITTER MENGGUNAKAN CONVOLUTIONAL LONG SHORT TERM MEMORY (CLSTM)

Anggraeni, Sherly Rosa (2021) KLASIFIKASI JENIS PENGADUAN DAN VISUALISASI LOKASI DARI CRAWLING DATA TWITTER MENGGUNAKAN CONVOLUTIONAL LONG SHORT TERM MEMORY (CLSTM). Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 05111740000018-Undergraduate_Thesis.pdf] Text
05111740000018-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2023.

Download (3MB) | Request a copy

Abstract

Berkembang pesatnya dunia teknologi informasi memudahkan manusia untuk bertukar informasi dengan cepat dan mudah melalui media sosial. Salah satu media sosial yang sering
digunakan adalah twitter. Pengaduan seperti jalan rusak, internet lambat, pdam tidak mengalir, mati listrik, atau pengaduan lainnya seringkali tidak terdata dan terlewatkan yang dapat menimbulkan dampak negatif kepada masyarakat, perlu diketahui sedini mungkin oleh masyarakat umum agar dapat terhindar dari dampak pengaduan. Sistem ini dibuat untuk memfasilitasi masyarakat dan pihak berwenang dalam penyebaran informasi terkait pengaduan secara daring.
Tugas akhir ini menggunakan metode Convolutional Long Short Term Memory. Tahapan dari tugas akhir ini adalah crawling
data Twitter, pre-processing pada data crawling, pelabelan
manual, pembentukan model, stream data Twitter, pre-processing pada data stream, klasifikasi secara real time, pengenalan entitas, pengenalan relasi, pencarian latitude dan longitude. Hasil dari jenis pengaduan divisualisasikan dalam bentuk web menggunakan framework Laravel dan bantuan API GoogleMaps.Hasil dari uji coba menunjukkan bahwa rata rata perhitungan precision, recall dan f-measure untuk pengenalan entitas sebesar 92,03%, 94,07%, 92,35%, untuk klasifikasi jenis pengaduan sebesar 90,11%, 89,66%, 89,85%, dan untuk pengenalan relasi sebesar 96,29%, 100%, 97,91%.
======================================================================================================
The rapid development of the world of information technology makes it easier for humans to exchange information quickly and easily via social media. One of the most popular social
media used is twitter. Complaints such as damaged roads, low
internet, no running water supply, power outages, or other
complaints are often not recorded and missed which can have a negative impact on the community, it is necessary to know as early as possible by the general public in order to avoid the impact of complaints. This system was created to facilitate the public and the authorities in disseminating information related to online complaints.
This final project uses the Convolutional Long Short Term
Memory method. The stages of this final project are Twitter data crawling, pre-processing on data crawling, manual labeling, model formation, Twitter data streams, pre-processing on data streams, classification in real time, entity recognition, relationship recognition, search for latitude and longitude. The results of this type of complaint are visualized in web form using the Laravel
framework and the help of the Google Maps API.
The results of the trial show that the average precision,
recall and f-measure calculations for entity identification are 92.03%, 94.07%, 92.35%, for classification types of complaints are 90.11%, 89.66%, 89.85%, and for relationship recognition 96.29%, 100%, 97.91%.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: Deteksi Pengaduan, Media Sosial, Aplikasi Web, Convolutional Long Short Term Memory, Complaint Detection, Social Media, Web Application
Subjects: T Technology > T Technology (General) > T57.5 Data Processing
T Technology > T Technology (General) > T58.6 Management information systems
Divisions: Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User: Sherly Rosa Anggraeni
Date Deposited: 31 Jul 2021 16:05
Last Modified: 31 Jul 2021 16:07
URI: http://repository.its.ac.id/id/eprint/84632

Actions (login required)

View Item View Item