Sentiment Classification

Sentiment analysis is widely used in several application domains, such as text mining, profile generation, movie recommendation and musical mood classification. In the state-of-the-art studies, the data type oof the query and search datasets are mostly uni-modal. In other words, either text, visual or audio data are processed for sentimental information extraction and classification. In our studies, we try to combine optical face images, search text and neural data acquired using functional magnetic resonance imaging (fMRI), for sentiment (such as happy, sad, satisfied, frustrated etc) classification using information fusion and machine learning techniques.

The ultimate goal of the project is to employ the sentiment class labels to improvbe the internet search engines of Google. Using the results of the sentiment classifiers one can identify and manipulate various characteristics of search engines.

Recent talk by Prof. Vural in Brain Awareness Week (in Turkish):


  • We thank National Magnetic Resonance Research Center (UMRAM) for their hospitality and support during fMRI experiments.
  • We also acknowledge Google for their support between 2011-2012, Scientific and Technological Research Council of Turkey (TUBITAK) and Coordinatorship of Scientific Research Projects, METU (BAP) for their ongoing supports since 2013.