Computational humor recognition is considered one of the most challenging tasks in Natural Language Processing (NLP) primarily due to the intricate nature of humor as an emotion. Although most studies on humor recognition have focused on English textual sources, much work has been done in other languages as well. However, there is a notable gap in the literature concerning the Greek language. This paper introduces the first-ever Greek Humorous Dataset (GHD), specifically designed to address this void in the literature. GHD is a manually annotated balanced dataset consisting of 10,000 short text samples labeled as either humorous or non-humorous. In addition to a detailed description of the dataset, we compare the performance of ten machine learning models using text representation feature engineering techniques to establish benchmarks for future research. With the development of GHD, we aim to not only contribute to the expanding field of knowledge in computational humor recognition but also foster a positive impact on future research endeavors in Greek language processing.