Şirketlere ilişkin haber metinleri ile finansal değerler arasındaki ilişkilerin nesnel bir temelde analiz edilebilmesini sağlamak için, öncelikle ilişkili haberlerin ve bu haberlerde yer alan olumlu ya da olumsuz ifadelerin sayısal değerlere dönüştürülmesi gerekmektedir. Bu amaçla literatürde "metin madenciliği" ve "Duygu (Sentiment) Analizi" yaklaşım ve yöntemleri kullanılmaktadır. Bu çalışmada da, Borsa İstanbul'da işlem gören BIST30 şirketlerine ilişkin 2014 yılında farklı kaynaklarda yayınlanmış 14.108 haber metin madenciliği teknikleri ile alınarak, yıllık ve çeyreklik bazda haber sayıları tespit edilmiştir. Haber içeriklerinde yer alan ifadeler de, Türkçe diline çevrilerek oluşturulmuş bir "Duygu Sözlüğü" yardımıyla, sayısal değerlere dönüştürülmüştür. Daha sonra, bu sayısal skorlar ile aynı dönemde piyasada oluşan şirket değerleri arasındaki ilişkiler analiz edilmiştir. Ortaya çıkan temel sonuç, finansal piyasalarla yayınlanan haberler ve bunların duygu tonları ile finansal değerler arasında anlamlı ilişkilerin var olduğudur. Bu sonuç Türk finansal piyasalarının değerlendirilmesinde önemli bir araç olarak Türkçe haber kaynaklarının da kullanılabileceğini ortaya koymaktadır. AbstractIn order to analyze the relations between financial values and the textual news related with the companies, there is a need for a conversion of textual content to quantitative values. For this aim, "text mining" approaches and "sentiment analysis" tools are used. In this framework, this study evaluates the relations between market capitalization of companies in Borsa Istanbul (BIST30) and published financial news about them. For this aim, published 14.108 news in the year 2014, have been extracted from 313 different news content providers and have been quantified by using text mining approaches. Then, for the statements in the news, a sentiment analysis has been performed by using a sentiment dictionary which is translated into Turkish in this study. Finally, the relations between these quantities and market capitalization values of the companies have been examined. The result of the analyses is the existence of relations between financial values of and published financial news about the companies. This study, via providing a methodological framework may help and show that the sources of financial news in Turkish also can be used as a novel tool for the analysis of Turkish financial markets.

q-Pareto rule, Sequential choice models, Two-stage, choice model, Criterial extremization, Reducibility,

Efficiency measurement is an important analysis for institutions in order to investigate their performance comparatively. On the other hand, evaluating efficiency of institutions carrying out more than one activity using common resources is important management problems. Since both efficiency and priorities may change in time, the problem set is getting more complex.Sub-activities of total system (Decision Making Sub-Units/DMSU) of institutions may act in different ways since they have different priorities or goals. In addition, factors effecting efficiencies of these DMSUs may differ. Each DMSUs' countermeasures can vary in order to reach the efficient frontier. Therefore, total system efficiency results may not represent the DMSUs' efficiencies.In line with these inferences, the aim of this study is to analyze efficiency differences between DMUs and DMSUs in terms of their behaviors, efficiencies and countermeasures in a dynamic manner. A hybrid non-parametric dynamic efficiency evaluation model, "Multi-Activity Window Data Envelopment Analysis (MA-WDEA)", is developed first[1]. Then, internal dynamics of DMUs and DMSUs are analyzed via dynamic returns to scale(RTS) analysis and Tobit regression models. Both analysis are established on efficiency results of the MA-WDEA model.The proposed process is applied to measure the performance of Turkish banking system. The results first show that MA-WDEA is a suitable tool to measure efficiency trends of DMUs/DMSUs. There exist important differences among Turkish banks with respect to their operating and non-operating activities’ efficiencies. The results of the dynamic RTS and regression models also show that behavioral attitudes of DMUs and DMSUs are significantly different.[1] The developed hybrid model is first presented in 3rd International Conference on Governance, Management and Entrepreneurship in Crotia by Kaya and Cinar, (2015).

No abstract

No abstract