Büyük Veri Metin Madenciliği Bağlamında İstanbul Büyükşehir Belediyesi Seçimlerinde Aday Liderlerin Tweet Kullanımları Üzerine Bir İnceleme
Yeşim Güçdemir, Seda Mengü, Kemal GünaySosyal medya bireylerin bilgilerini, fikirlerini, deneyimlerini paylaştığı ve bunları başkalarına aktardıkları platformlardır. Siyasi aktörler de hedef gruplarının bu mecralarda yer alması nedeniyle bu platformlarda seçmen kitlelerini artırmaya, siyasete katılımı sağlamaya yönelik çalışmalar yapmaktadırlar. Özellikle genç kuşakların televizyon, gazete gibi geleneksel medyalar yerine sosyal medya uygulamalarını kullanmaları siyasi aktörler için sosyal ağları daha da önemli kılmaktadır. Metin madenciliği ile bu ortamlarda farklı biçimlerde yer alan büyük miktardaki verinin analiz edilmesi ve anlamlandırılması mümkün olmaktadır. Bu çalışmada İstanbul Büyükşehir 23 Haziran 2019 seçimlerinde başı çeken iki aday Binali Yıldırım ve Ekrem İmamoğlu’nun yürütmüş olduğu siyasal iletişim stratejilerinde bir sosyal ağ uygulaması olan Twitter üzerinde söylemleri üzerinden liderlik özelliklerinin belirlenmesi ve başarı sağlayan liderin izlediği stratejideki hangi liderlik özelliklerini söylemlerinde kullandığı ve bunların siyasal iletişim bağlamında yansıması analiz edilmiştir. Çalışmada yöntem olarak betimsel araştırma modeli (Yıldırım & Şimşek, 2018, p. 239) karşılaştırma yaklaşımı esas alınmıştır. Nitel ve nicel yöntemler bir arada kullanılarak karma araştırma desenleri de araştırma içerisinde yer almaktadır. Aynı zamanda her iki adayın tweetleri üzerinden metin madenciliği yöntemleri kullanılarak söylem ve duygu analizleri gerçekleştirilmiştir. Yapılan “İddialılık-Duyarlılık Ölçümü” sonucunda Binali Yıldırım’ın literatürde bahsedilen “Başarı Odaklı Liderlik” özelliklerini gösterdiği, Ekrem İmamoğlu’nun ise “Destekleyici Liderlik” özelliklerini gösterdiği saptanmıştır. Liderlerin Twitter üzerindeki paylaşımlarından yola çıkılarak yapılan “İddialılık-Duyarlılık Ölçümü” sonucunda Ekrem İmamoğlu’nun her iki kategoride de Binali Yıldırım’a göre daha iddialı ve duyarlı olduğu sonucu ortaya çıkmıştır. Liderlerin sözlük tabanlı olarak yapılan duygu analizinde ise anlamlı bir farklılık olmadığını tespit edilmiştir.
An Investigation of Candidate Leaders’ Tweet Campaigns Prior to the Istanbul Metropolitan Municipal Elections Using Big Data Text Mining
Yeşim Güçdemir, Seda Mengü, Kemal GünaySocial media provide platforms where individuals share their knowledge, ideas, and experiences and then transfer them to others. Political actors are also working to increase their access to the electoral masses and to participate in politics on these platforms as the target groups are included in these channels. Younger generations’ use of social media rather than traditional media such as television and newspapers make social networks more important for political actors. Using text mining, it is possible to analyze and make sense of a large amount of data in different formats in these environments. In this study, their leadership characteristics of two political candidates, Binali Yıldırım and Ekrem İmamoğlu, who were the leaders of the Istanbul Metropolitan 23 June 2019 elections, were determined by examining their discourses on Twitter. The leadership features of the discourse strategy followed by the successful leader and their political supporters were analyzed. The study deployed a descriptive research model (Yıldırım & Şimşek, 2018, p. 239) and a comparative method. A mixed research design was used which included both qualitative and quantitative methods. At the same time, discourse and emotion analyzes were conducted using text mining methods on the tweets of both candidates. The result of the “assertiveness-sensitivity measurement” indicated that Binali Yıldırım demonstrated “successoriented leadership” features, and Ekrem İmamoğlu demonstrated “supportive leadership” features. The result of the “assertiveness-sensitivity measurement” made based on the leaders’ Twitter feed indicated that Ekrem İmamoğlu was both more assertive and sensitive than Binali Yıldırım. There was no significant difference in the dictionary-based analysis of the emotions expressed by leaders.
As a result of the digitalization of every area of life, individuals’ work practices and lifestyles have changed. Today, this process has accelerated with the internet of things, and design innovations with digital and smart production technologies. The development of internet and mobile technology infrastructures, the increase in digitalization, sensor technologies, portability, and mobility have become more important as data processing based on communication technologies takes up more space in our lives with the widespread use of mobile web pages.
This study aimed at identifying the leadership characteristics of two political candidates, Binali Yıldırım and Ekrem İmamoğlu, as leaders of the 2018 June Istanbul Metropolitan Municipality Elections, by analyzing their discourses on a social network application in a political communication context. To this end, discourse and emotion analyzes were conducted on both candidates’ tweets using data mining techniques.
This research framework investigated the following three issues. First, it identified the leadership characteristics of both political leaders and which leadership characteristics came to the fore during their tweets. Second, the discourses that emerged during the posts of political leaders were identified. Third, the dominant emotional variables were identified by conducting a textual analysis of the emotional content of candidates’ discourses.
The study deployed a descriptive research model (Yıldırım & Şimşek, 2018, p. 239) and a comparative method. A mixed research design was used which included both qualitative and quantitative methods. A Twitter Developer Account was created to obtain the data. An API (Application Programming Interface) was taken to pull data from Twitter. All tweets were obtained using the RStudio program via the API. Tweets were converted into data frame/tibble data format by the program in preparation for analysis.
The “assertiveness-responsiveness measurement” of Richmond and McCroskey (Richmond & McCroskey, 1990) was used to analyze leadership characteristics. Following this scale, a coding category table was created on the NVivo program for leadership studies. Assertiveness items included defends own beliefs, independent, forceful, has a strong personality, assertive, dominant, willing to take a stand, acts as a leader, aggressive, and competitive. Responsiveness items included helpful, responsive to others, sympathetic, compassionate, sensitive to the needs of others, sincere, gentle, warm, tender, and friendly.
Since the Istanbul Büyükşehir (IBB) Elections were held between the two candidates, a purposive sampling method was used. “Purposeful sampling allows situations that are thought to have rich information to work in-depth” (Yıldırım & Şimşek, 2018, p. 118). A different data set was created for the “assertiveness and sensitivity measurement” of the study. For each candidate, the top 100 tweets with the most user interaction were identified and coded. During the discourse and emotion analysis sections, all the candidates’ tweets were taken as samples, 602 tweets for Binali Yıldırım and 957 tweets for Ekrem İmamoğlu.
This study identified the leadership characteristics of both candidates through an examination of the discourses of both political leaders. The result of the “assertivenesssensitivity measurement” indicated that Binali Yıldırım demonstrated “success-oriented leadership” features, and Ekrem İmamoğlu demonstrated “supportive leadership” features. The result of the “assertiveness-sensitivity measurement” made based on the leaders’ Twitter feed indicated that Ekrem İmamoğlu was both more assertive and sensitive than Binali Yıldırım. There was no significant difference in the dictionary-based analysis of the emotions expressed by leaders.
The data was accessed to examine the leaders’ Twitter discourses. Non-structured data can be processed and transformed into a structured order. By using data visualization techniques, the frequencies of the leaders’ expressions, sub-patterns in the candidates’ messages and emotional inferences were made based on a dictionary analysis. The combined application of these methods to a social network is an effective tool for monitoring people’s preferences, observing trends and making predictions about the future.
In this study, we applied data acquisition, cleaning data, analyzing data, data visualization steps to benefit from big data for academic purposes. The use of computerbased programming language and algorithm-supported research methods and techniques make this research significant in terms of communication sciences. Besides being a commercial commodity, big data can also be used in academic studies as in this study. In the field of communication sciences, more technical research can be done with machine learning and deep learning methods. As a result, this study will contribute to the gap in the Turkish literature regarding big data in the field of communication. It will contribute to communication sciences in terms of the methodology by using different research applications such as text processing, data visualization and dictionarybased emotion analysis.
