Developed a Python tool for retrieving documents from large text corpora.
Technical Stack: Python, Word2Vec, TF-IDF
Developed a GAN-based solution to simulate and detect fake responses in online surveys. Introduced a data-independent mechanism for fake response detection using overtrained discriminators.
Technical Stack: Python, Generative Adversarial Networks (GANs), Machine Learning.
Extracted emotional dimensions from Reddit text using advanced word embeddings. Revealed patterns and clusters in emotional vectors through PCA visualizations.
Technical Stack: Python, NLP, PCA, Word Embeddings (Doc2Vec)
Developed a state-of-the-art transformer-based neural network architecture for extrapolating semantic and emotional norms across multiple languages. Achieved superior correlations with human judgments, improving on prior models by Δr = 0.1 on average, and introduced a method for unsupervised control in stimuli selection. The project was accepted for publication in Behavior Research Methods.
Technical Stack: Python, Transformers, Machine Learning, Natural Language Processing.
Developed a Python implementation of the original R package statcheck by Michèle B. Nuijten, enhancing accessibility for the Python community. The tool automatically extracts NHST results from articles, recomputes p-values, and detects inconsistencies. Supports various statistical tests, including t-tests, F-tests, and z-tests, when reported in APA style. Main applications include manuscript self-checks, aiding peer review processes, and research, such as investigating predictors of statistical inconsistencies.
Technical Stack: Python, Text Processing, Statistical Analysis
Built an application for user-friendly exploratory bias detection in meta-analyses, which allows users to assess the reliability of scientific literature via keyword searches.
Technical Stack: R, Shiny, Meta-Analysis.