Text mining and practical content analysis applications in Turkish texts.

Öz

The most significant challenge in applying text mining to Turkish texts is that, due to the agglutinative nature of the language, preprocessing and normalization decisions directly affect the analysis results. Despite this, application guides that explain, along with their justifications, the steps involved in preparing data for Turkish texts, creating linguistic normalization text representations, extracting thematic patterns, and reporting outputs in R software are still limited. This study aims to address this deficiency by presenting a step-by-step workflow for generating text statistics and conducting content analysis in Turkish texts using R software. The material for this study consists of a structure of 30 Turkish texts produced on the same subject. In the method, indicators related to length, sentence structure, lexical diversity, conjunction, stop word ratios, and punctuation profile were first calculated for the raw texts to gather potential background information. Then, preprocessing and root reduction steps appropriate for Turkish were applied, document-term matrices were created, and a reasoned selection was made by comparing candidate solutions for thematic pattern extraction using multiple quality indicators. The findings revealed significant differences in length and thematic content density among the texts, and a decrease in the volume of tokens included in the analysis after preprocessing. The selected thematic structure was found to produce a moderate level of similarity in repeated runs. In conclusion, the study presents an application framework that reveals critical decision points in the text analysis process conducted with R software in Turkish texts and can produce similar outputs when repeated with the same data.

Anahtar Kelimeler

• text mining
• content analysis
• thematic patterns
• preprocessing
• r

Türkçe metinlerde metin madenciliği ile pratik içerik analizi uygulamaları

Öz

Türkçe metinlerde metin madenciliği uygulamalarında en önemli güçlük, dilin sondan eklemeli yapısı nedeniyle ön işleme ve normalleştirme kararlarının analiz sonuçlarını doğrudan etkilemesidir. Buna rağmen R yazılımında; Türkçe metinler için veri hazırlama, dilsel normalleştirme, metin temsili oluşturma, tematik örüntü çıkarımı ve çıktıların raporlanması gibi adımları birlikte ve gerekçeleriyle açıklayan uygulama kılavuzları hala sınırlı düzeydedir. Bu çalışma, bu eksikliği azaltmak amacıyla, R yazılımı kullanılarak, Türkçe metinlerde metin istatistikleri üretme ve içerik analizi yürütme sürecini adım adım bir iş akışı olarak sunmuştur. Çalışmanın materyalini, aynı konu ekseninde üretilmiş, 30 Türkçe metinden oluşan bir yapı oluşturmuştur. Yöntemde, öncelikle potansiyel ön bilgi için; ham metinler için uzunluk, cümle yapısı, sözcüksel çeşitlilik, bağlaç, durak sözcük oranları ve noktalama profiline ilişkin göstergeler hesaplanmıştır. Ardından Türkçeye uygun ön işleme ve köke indirgeme adımları uygulanmış, belge–terim matrisleri oluşturulmuş ve tematik örüntü çıkarımı için aday çözümlerin birden fazla kalite göstergesiyle karşılaştırılması sonucunda gerekçeli bir seçim yapılmıştır. Bulgular, metinler arasında belirgin uzunluk ve tematik içerik yoğunluğu farklılıkları bulunduğunu ve ön işleme sonrasında analize giren token hacminde azalma gerçekleştiğini göstermiştir. Seçilen tematik yapının tekrarlı çalıştırmalarda orta düzeyde benzerlik ürettiği görülmüştür. Sonuç olarak çalışma, Türkçe metinlerde R yazılımı ile yürütülen metin analizi sürecinde kritik karar noktalarını görünür kılan ve aynı veri ile tekrarlandığında benzer çıktılar üretebilen bir uygulama çerçevesi sunmuştur.

Anahtar Kelimeler

• metin madenciliği
• içerik analizi
• tematik örüntü
• ön işleme
• r

Kaynakça

1. Alkan, B. B., Karakuş, L., & Direkci, B. (2022). Generating a sentiment dictionary in R and dictionary-based sentiment analysis in Turkish texts. Digital Scholarship in the Humanities, 38(2), 461–476. https://doi.org/10.1093/llc/fqac093
2. Alkan, B. B., Karakuş, L., & Direkci, B. (2023). Knowledge discovery from the texts of Nobel Prize winners in literature: sentiment analysis and Latent Dirichlet Allocation. Scientometrics, 128(9), 5311–5334. https://doi.org/10.1007/s11192-023-04783-6
3. Belford, M., Mac Namee, B., & Greene, D. (2018). Stability of topic modeling via matrix factorization. Expert Systems with Applications, 91, 159–169. https://doi.org/10.1016/j.eswa.2017.08.047
4. Benoit, K., Watanabe, K., Wang, H. P., Nulty, P., Obeng, A., Müller, S., … & Matsuo, A. (2018). quanteda: An R package for the quantitative analysis of textual data. The Journal of Open Source Software, 3(30), 774. https://doi.org/10.21105/joss.00774
5. Benoit, K., Watanabe, K., Wang, H., Nulty, P., Obeng, A., Müller, S., & Matsuo, A. (2018). quanteda: An R package for the quantitative analysis of textual data. Journal of Open Source Software, 3(30), 774. https://doi.org/10.21105/joss.00774
6. Denny, M. J., & Spirling, A. (2018). Text preprocessing for unsupervised learning: Why it matters, when it misleads, and what to do about it. Political Analysis, 26(2), 168–189. https://doi.org/10.1017/pan.2017.44
7. Eryiğit, G., & Torunoğlu-Selamet, D. (2017). Social media text normalization for Turkish. Natural Language Engineering, 23(6), 835–875. https://doi.org/10.1017/s1351324917000134
8. Feinerer, I., Hornik, K., & Meyer, D. (2008). Text Mining Infrastructure in R. Journal of Statistical Software, 25(5). https://doi.org/10.18637/jss.v025.i05

9. Gągolewski, M. (2022). stringi: Fast and Portable Character String Processing in R. Journal of Statistical Software, 103(2). https://doi.org/10.18637/jss.v103.i02
10. Greene, D., O’Callaghan, D., & Cunningham, P. (2014). How many topics? Stability analysis for topic models. In Advances in Information Retrieval (ECIR 2014) (pp. 498–503). Springer. https://doi.org/10.1007/978-3-662-44848-9_32
11. Grimmer, J., & Stewart, B. (2013). Text as Data: The Promise and Pitfalls of Automatic Content Analysis Methods for Political Texts. Political Analysis, 21(3), 267–297. https://doi.org/10.1093/pan/mps028
12. Hannigan, T. R., Haans, R. F. J., Vakili, K., Tchalian, H., Glaser, V., Wang, M. S., … & Jennings, P. D. (2019). Topic Modeling in Management Research: Rendering New Theory from Textual Data. Academy of Management Annals, 13(2), 586–632. https://doi.org/10.5465/annals.2017.0099
13. Hubert, L. J., & Arabie, P. (1985). Comparing partitions. Journal of Classification, 2(1), 193–218. https://doi.org/10.1007/bf01908075
14. Jaccard, P. (1901). Étude comparative de la distribution florale dans une portion des Alpes et du Jura. Bulletin de la Société Vaudoise des Sciences Naturelles, 37, 547–579. https://doi.org/10.5169/seals-266450
15. Kılınç, D., Özçift, A., Bozyiğit, F., Taşer, P. Y., Yücalar, F., & Borandağ, E. (2015). TTC-3600: A new benchmark dataset for Turkish text categorization. Journal of Information Science, 43(2), 174–185. https://doi.org/10.1177/0165551515620551
16. Mäntylä, M. V., Claes, M., & Farooq, U. (2018). Measuring LDA topic stability from clusters of replicated runs. In Proceedings of the 12th ACM/IEEE International Symposium on Empirical Software Engineering and Measurement (ESEM ’18). ACM. https://doi.org/10.1145/3239235.3267435
17. Mimno, D., Wallach, H. M., Talley, E., Leenders, M., & McCallum, A. (2011). Optimizing semantic coherence in topic models. In Proceedings of the 2011 Conference on Empirical Methods in Natural Language Processing (pp. 262–272). Association for Computational Linguistics. https://aclanthology.org/D11-1024/
18. Munafò, M. R., Nosek, B. A., Bishop, D., Button, K. S., Chambers, C., Sert, N. P. d., … & Ioannidis, J. P. A. (2017). A manifesto for reproducible science. Nature Human Behaviour, 1(1). https://doi.org/10.1038/s41562-016-0021
19. O’Callaghan, D., Greene, D., Carthy, J., & Cunningham, P. (2015). An analysis of the coherence of descriptors in topic modeling. Expert Systems With Applications, 42(13), 5645–5657. https://doi.org/10.1016/j.eswa.2015.02.055
20. Oflazer, K. (1995). Two-level Description of Turkish Morphology. Literary and Linguistic Computing, 9(2), 137–148. https://doi.org/10.1093/llc/9.2.137
21. Peng, R. D. (2011). Reproducible Research in Computational Science. Science, 334(6060), 1226–1227. https://doi.org/10.1126/science.1213847
22. Röder, M., Both, A., & Hinneburg, A. (2015). Exploring the space of topic coherence measures. In Proceedings of the Eighth ACM International Conference on Web Search and Data Mining (WSDM ’15) (pp. 399–408). Association for Computing Machinery. https://doi.org/10.1145/2684822.2685324
23. Sebastiani, F. (2002). Machine learning in automated text categorization. ACM Computing Surveys, 34(1), 1–47. https://doi.org/10.1145/505282.505283
24. Silge, J., & Robinson, D. M. (2016). tidytext: Text Mining and Analysis Using Tidy Data Principles in R. The Journal of Open Source Software, 1(3), 37. https://doi.org/10.21105/joss.00037
25. Torunoğlu, D., Cakirman, E., Ganiz, M. C., Akyokuş, S., & Gütbüz, M. Z. (2011). Analysis of preprocessing methods on classification of Turkish texts. In 2011 International Symposium on Innovations in Intelligent Systems and Applications (pp. 112–117). https://doi.org/10.1109/inista.2011.5946084
26. Yıldız, O. T., Avar, B., & Ercan, G. (2019). An open, extendible, and fast Turkish morphological analyzer. In Proceedings of the International Conference on Turkish Natural Language Processing (RANLP/ACL Anthology). https://aclanthology.org/R19-1156/.