PhD colloquium: Emotion and Sentiment Detection in Unstructured Social Data

Vortragender: Maha Alghalibi

Datum & Uhrzeit: 13.11.2019, 16:15 - 17:45 Uhr, Gebäude D 239

Gastgeber: J.-Prof. Kai Lawonn, Prof. Dietrich Paulus

Titel: Emotion and Sentiment Detection in Unstructured Social Data


During the relatively ‘short’ period of time elapsed between the last thirty years of the 20th century and the end of the first two decades of this 21st century, the world has progressively witnessed an ‘explosive’ advancement in the diverse multidisciplinary fields of computer science/engineering, communication engineering, and information technology. However, increasing the size of data, especially in social media platforms, led to the emergence of many challenges. One of these challenges is how to extract knowledge from a big chunk of data which would result in finding useful information for easily understand, interpret and make a decision.

We address the problem of detection, classification and quantification of emotions and opinion of text and image in any form. We consider English text collected from social media like Twitter and dataset of images polarity from Flicker and twitter which can provide information having utility in a variety of ways, especially sentiment, emotion and opinion mining. Social media like Twitter, Facebook, and Flicker are full of emotions, feelings and opinions of people all over the world. However, analyzing and classifying text and image on the basis of emotions and sentiment is a big challenge and can be considered as an advanced form of Sentiment Analysis. We propose a method to classify text and image into different Emotion-Categories: Negative, Positive and Neutral. In our proposed system, we use two different approaches and combine them to effectively extract these emotions from text and image. The first approach is based on Natural Language Processing, and uses several textual features and dimensionality reduction methods. The second approach is based on Machine Learning classification algorithms. On testing, it is shown that our model provides significant accuracy in classifying tweets and images taken from Twitter and Flicker.

Wann 13.11.2019
von 16:15 bis 17:45
Termin übernehmen vCal