Twitter Information Extraction for Smart City

Abstract

Abstrak-In Indonesia, Bandung is the second most activeTwitter user, which means a lot of tweets have been shared among Bandung people on Twitter. Tweets can be used as a data source to explore information related to the city. One example is information related to traffic congestion, such as information of location, date, and time when the traffic congestion happened. In this study, we proposed a method to filter the tweets related to traffic congestion in Bandung and to extract the information of location, time, date and image (if any). SVM with several variations of the weighting and the selection of features is used for filtering process. The results showed that the greatest accuracy rate is 83% use Binary weighting method in top-2000 features. Meanwhile, information extraction process carried out by a rule-based approach, gave satisfactory results, around 98% -100% for the extraction of date, time and URL. However, the extraction of location information only gave accuracy of about 62%. It was caused by OOV (Out of Vocabulary) and OOR (Out of Rules).

We Know You Are Living in Bali: Location Prediction of Twitter Users Using BERT Language Model

“…Twitter provides an application programming interface (API) https://developer.twitter.com/en/docs/twitter-api (accessed on 17 May 2022) enables the collection of data from the site using its web service. With the help of Twitter API, tweet data can be analyzed for various tasks and applications, including sentiment and stance analysis [3][4][5][6][7][8], hate speech and misinformation detection [9][10][11][12][13][14], traffic monitoring [15][16][17], disaster management [18][19][20], and disease outbreak control [21,22].…”

A study on extracting disaster information from tweets

“…There are many existing studies on classifying huge amount of data in tweets: a study to extract events from twitter [Ritter et al, 2010]; a study to extract operation status of railroad systems [Tsuchiya et al, 2013]; studies to extract road traffic information [Hanifah et al, 2014;Sakaki et al, 2015]; and a study to extract tourist information [Obara et al, 2015]. As for the classification of disaster tweets, there is a study by Rokuse [Rokuse et al, 2015].…”

Supervised Entity Tagger for Indonesian Labor Strike Tweets using Oversampling Technique and Low Resource Features

“…Furthermore, Indonesian social media data particularly exhibits the mixed use of languages including the official language Bahasa Indonesia, English and several Indonesian regional languages. For Indonesian social media, there are several researchers on NER ( [1][2][3][4][5]). Existing approaches can generally be divided into rule-based ( [1,2]) and statistical approaches ( [3][4][5]).…”

“…For Indonesian social media, there are several researchers on NER ( [1][2][3][4][5]). Existing approaches can generally be divided into rule-based ( [1,2]) and statistical approaches ( [3][4][5]). In rule based systems, researchers define rules in from of string patterns used to identify and classify named entities.…”

“…Existing studies applied Indonesian NER for different applications scenarios in multiple specific domains including e-commerce transactions ( [3]), citizen complaints ( [4]) and traffic conditions ( [1][2], [5]); and generic domains ( [6,7]). We developed the first entity tagger for tweets sent during civic strike and protest events in Indonesia.…”