Ensemble Methods for Personality Recognition

An important bottleneck in the development of accurate and robust personality recognition systems based on supervised machine learning, is the limited availability of training data, and the high cost involved in collecting it. In this paper, we report on a proof of concept of using ensemble learning as a way to alleviate the data acquisition problem. The approach allows the use of information from datasets from different genres, personality classification systems and even different languages in the construction of a classifier, thereby improving its performance. In the exploratory research described here, we indeed observe the expected positive effects.