LINGUISTIC STUDIES ON TWEETS GATHERED FROM MUĞLA REGION: A PRELIMINARY STUDY

YÖNETİM BİLİŞİM SİSTEMLERİ DERGİSİ http://dergipark.ulakbim.gov.tr/ybs/ Yayın Geliş Tarihi: 30.05.2016 Cilt:1, Sayı:3, Yıl:2016, Sayfa 29-39 Yayına Kabul Tarihi: 15.08.2016 ISSN: 2148-3752 Online Yayın Tarihi: 05.10.2016 LINGUISTIC STUDIES ON TWEETS GATHERED FROM MUĞLA REGION: A PRELIMINARY STUDY Feriştah Dalkılıç Enis Karaarslan Ali Hurriyetoglu Abstract Citizens or visitors of a city can supply significant information with their social media posts by using mobile devices. These data can give information about complaints, touristic attractions, emergency situations etc. Social media analysis will be beneficial for smart city and smart management concept. This study is a first attempt to analyze and understand this touristic Muğla region by using social media. During this study, a sample dataset is formed by collecting the tweets that were sent from the Muğla region. Linguistic studies are implemented in tweets which are in Turkish language. Various techniques, statistical language and characteristics are used. Preliminary study revealed main topics about the region, user and hashtag types. We consider this analysis as a first step to a more detailed and complete study for this region. Keywords: social media analysis, tweet, smart city, crowd sensing MUĞLA BÖLGESINDEN TOPLANAN TWEET VERILERI ÜZERINDE DILBILIMSEL ÇALIŞMALAR: ÖN ÇALIŞMA Öz Bir şehrin sakinleri ve ya ziyaretçileri, mobil cihazlar aracılığıyla sosyal ağlar üzerinde önemli miktarda veri üretmektedirler. Bu veriler; şikâyetler, acil durum halleri, turistik eğlence programları gibi konularda bilgi sağlayabilmektedir. Sosyal medya analizinin, akıllı şehir ve akıllı yönetim sistemleri için faydalı olacağı düşünülmektedir. Bu çalışma, turistik bir bölge olan Muğla yöresini sosyal medya kullanarak analiz etmek için bir ilk girişimdir. Bu çalışma süresince, Muğla yöresinden atılan tweetler toplanarak örnek bir veri seti oluşturulmuştur. Türkçe tweetler üzerine dilbilimsel bir çalışma uygulanmıştır. Çeşitli teknikler, istatistiksel dil ve karakteristikler kullanılmıştır. Ön çalışma, sosyal ağda kullanılan dilin özelliklerini, yöre hakkındaki ana konuları, kullanıcı ve hashtag tiplerini ortaya sermiştir. Bu çalışma, yöre hakkında daha ayrıntılı ve tam bir analiz için ilk adımı oluşturmaktadır. Anahtar Kelimeler: sosyal medya analizi, tweet, akıllı şehir, kalabalık algılama Dalkılıç, F., Karaarslan, E., Hürriyetoğlu, A. Yönetim Bilişim Sistemleri Dergisi, Cilt:1, Sayı:3 INTRODUCTION Smart city concept is built on smart management of all city structures like power, water, and transportation etc. for a safe, secure and efficient usage of resources. Several technologies can be used for this concept such as sensors, electronics, computerized systems, networks and communication systems etc. (Bowerman et al., 2000). We are living in a knowledge society. Citizens or visitors of a city share the up-to-date events and their emotions to these events by using social media (Twitter, Facebook, Instagram etc.) environments. These environments have publicly open, free and real time data. These data can be analyzed to obtain complaints, touristic attractions and emergency situations of the cities. The produced knowledge can be used for smart management of the cities. Analysis of social media data and converting it to benefit is one the most important research topics today. The big data and cloud environments are becoming widespread everyday facilitated by the fact that these can be set up with a less economical cost than before. These systems make social media data processing and storage easier. Social media analysis is used for journalism, tourism and commercial applications such as public research about the companies. It’s also possible to use it for scientific research and the public interest purposes such as collecting information about natural disasters or national security. Retrospective and instantaneous detection can be done with social media analysis; also predictions about the future can be made in light of the information obtained. Starting with a small dataset, this research aims to reach some preliminary results which will be used in the following studies. In the first section, related work will be given. Next, the usage of Social Media for Smart City and Smart Management is discussed. Then the implementation and experimental results will be given. Lastly, detected results and the possible future work will be discussed. RELATED WORK Many studies have been developed using Twitter data for a variety of purposes. Most of the studies are on sentiment analysis. Mohammad and Kiritchenko (2015) used hashtags to capture fine emotion categories from tweets. Cavazos-Rehg et al. (2016) examined depressionrelated content in Twitter to glean insight into social networking about mental health. Kunneman et al. (2014) experimented the majority-based and machine-learning methods for the identification of future event start dates from Twitter streams. Serrano et al. (2015) presented a simulation tool implementing several models of rumor diffusion in Twitter. Shendge et al. (2015) worked on real time Tweet analysis for event detection and reporting system for Earthquake. Tumasjan et al. (2010) presented a study uses the context of the German federal election to investigate whether Twitter is used as a forum for political deliberation and whether online messages on Twitter validly mirror offline political sentiment. Understanding a city through lens of social media has been studied by Bakıcı et al. (2013). Social event detection (Ilina et al., 2012), traffic event detection (Anantharam et al., 2015), touristic support (Leung et al., 2013; Quercia et al., 2014) are the most prominent recent approaches. Every bit of information in the aforementioned lines and the ones yet to be discovered contributes to understanding and management of a city. Social media analysis can also be used for crowd sensing and some implementation examples are given in some studies like (Roitman et al., 2012). A recent work (Preece et al., 2015) proposes a method to get active response from the tweeters. Multilingual Analysis of twitter data is important and as far as our knowledge, there are not many academic studies which focus in analyzing tweets in Turkish language. In an 30 Dalkılıç, F., Karaarslan, E., Hürriyetoğlu, A. Yönetim Bilişim Sistemleri Dergisi, Cilt:1, Sayı:3 interesting study (Zielinski, 2012), multilingual (Romanian, Greek and Turkish) analysis of Twitter data is implemented to detect human responses during earthquakes. In a recent study (Demirci, 2014), micro-blog entries and special usage of symbols and conveniences are studied. This work states that a new data set of Turkish tweets for emotion analysis is constructed. SOCIAL MEDIA ANALYSIS FOR THE SMART CITY Smart city and smart management depends upon collecting real-time data. Common interest of the citizens can be learned by collecting data from the sensing devices which is ca (...truncated)