GRAFFAIN is a Python Shiny supported multi-module online analysis software that allows qualitative data to be analyzed with the help of large language models and the findings to be visualized.

Although GRAFFAIN aims to increase the reproducibility of qualitative data analyses, it does not reject the differentiation arising from the researcher's experiences and related theories due to the nature of qualitative data analyses. Therefore, GRAFFAIN's basic philosophy is not to replace researchers but to assist them.

GRAFFAIN uses multilingual pre-trained language models while analyzing data. In this respect, the power of the analyses presented and the analysis performance of the modules are limited by the capacity of the language model used in that module. Although GRAFFAIN aims for language models to consistently display their capabilities, it accepts parsing/classification/comparison errors that may occur. However, GRAFFAIN aims to be a constantly developing and changing platform and to minimize errors. The amount of errors that may occur may vary depending on the data being in different languages. Researchers are advised to compare their findings with the formats of their datasets in different languages, taking this into account.

Different modules in GRAFFAIN are designed for different purposes and the accessed language models are trained with different prompts and different datasets in accordance with the purpose of the module. The performance of the module may increase or decrease in proportion to the similarity between the datasets used for pre-training and the user's dataset. The amount of error that will occur may also increase due to differences between the model's training purpose and the researcher's usage purpose, as well as the capacity of the language model.

GRAFFAIN modules are trained for specific purposes and with limited datasets, and do not provide effective results in unintended uses. Users are advised to check the results they receive from GRAFFAIN in detail, taking all these into account.

The technical report containing the detailed technical background for each module will be published in the documents section of our website shortly. You can access the mini guide prepared for the use of the modules by clicking the button below:

GRAFFAIN is a free platform developed independently and without support as a result of the dedicated efforts of developers to contribute to scientific & social research. If you use GRAFFAIN in your studies, citing the developers is the only expectation in return for this effort.

Citation:

Çüm, S., Demir, E. K., Demir, T., & Kahyaoğlu Erdoğmuş, Y. (2025). GRAFFAIN: Mini Guide [Preprint]. figshare. https://doi.org/10.6084/m9.figshare.30810395

GRAFFAIN founding and executive team thought and acted together in idea, design, creation and test processes apart from the contributions mentioned below. It will continue on its way with the same understanding in the process of continuously developing GRAFFAIN with user feedback.

GRAFFAIN is open to researchers' contributions in every sense. In this sense, you can contact the developer of the module you want to provide feedback directly from the contact addresses below. GRAFFAIN's modular structure reflects a philosophy parallel to its continuous growth goals. Existing modules will be constantly developed and new modules will be added. GRAFFAIN executive team extends an open invitation to all researchers who want to work together and develop new modules. Researchers who want to contribute will take their place on the platform with the module they developed as part of the GRAFFAIN volunteer team.

Assoc. Prof. Dr. Sait Çüm

Dokuz Eylül University Measurement and Evaluation in Education

Contact: sait.cum@deu.edu.tr
Modules Written: Content Analysis, Network Analysis, Sentiment Change Analysis, Mainstream & Marginal View Analysis, Shiny Design, Shiny UI & Server

Assoc. Prof. Dr. Elif Kübra Demir

Ege University Measurement and Evaluation in Education

Contact: elif.kubra.demir@ege.edu.tr
Modules Written: Network Analysis, Emotion Analysis, Risk Matrix, Shiny UI & Server

Dr. Tolga Demir

Ministry of National Education Innovation and Educational Technologies

Contact: demir.tolga@yahoo.com
Modules Written: Content Analysis, Needs Analysis, Political Axis Analysis, Shiny UI & Server

Asst. Prof. Dr. Yasemin Kahyaoğlu Erdoğmuş

Dokuz Eylül University Computer and Instructional Technologies Education

Contact: yasemin.kahyaoglu@deu.edu.tr
Modules Written: Logo Design, Website Design, Web Hosting, Website Frontend and Backend