In our research we addressed the problem of emotion elicitation of social networks. People tend to use social networks as a mean to express their emotions. Some people for example use a social network like Facebook as an outlet to express their negative emotions like frustration or anger and expect their network of friends to interact with them in a way that might help them deal with such emotions. Through friends' comments and likes people get feedback and support that could help neutralize their negative emotions. In the same manner, people use such networks to express their positive emotions like happiness and excitement. They expect their network of friends to interact with them through comments and likes which usually leads to the amplification of such emotions. Our aim in our research is to study emotion elicitation from social networks interactions. We studied Facebook data; we mainly focused on status updates and the comments and likes on those updates. We devised various experiments that helped use understand more the effect of such social interaction on people's emotion. We started our research work by reviewing the work done in the literature. We found out that researches have purely dealt with the emotion elicitation problem of social networking from a very limited perspective, and only used the textual features of social networks Ã¢â‚¬Å½Ã¢â‚¬Å½. They ignored the wealth of the other sources of information that could be used to better detect emotions of the users of the social networks, such as social graph of participants, preferences, location, comments, likes, events, images, audio and much more. We conducted a survey to better understand the social networks features that affect the emotions of users of Facebook. We found out that Facebook users tend to express their emotions using status, comments and likes features more than any other features provided by the social network. The emotions of the Facebook users are positively affected when the numbers of likes to their posts increase. The degree of connection of the person commenting or liking the users' comments makes a difference in the way the users are affected. Users tend to be affected with the posts of close friends the most, then comes the family members after. Gifts and events features of Facebook do not have a great impact on the emotions of Facebook users. Our findings guided us to focus on the likes, comments and degree of connection and the relationships between users and their friends in our study of the use of the various social network features to be used in emotion elicitation. We started to experiment with ConceptNet GuessMood function and we found that it is limited to only the six basic Ekman emotions which are happy, surprised, sad, angry, disgusted and fearful. Combining different approaches of emotion detection, we managed to expand the six basic emotions to a new set of 18 emotions. We ran an experiment using a labeled dataset to measure the accuracy of the expanded emotions. This hybrid technique resulted in an accuracy of 64.39 %. After expanding the six basic emotions, we researched the various features of Facebook that affects people emotions namely the likes, comments and the degree of connection. We started with the impact of receiving likes on the status message only on the emotions of the social network users. We aimed at identifying the weight of likes which maximizes the accuracy of the correctly detected emotions. We found out that assigning the likes the weight of 0.2 achieved the best accuracy which is 70.56%. We investigated the impact of the likes of the status message and relationships of the friends who made the likes. We found out that the incorporating different weights to likes by different friend relationships leads to an accuracy of 73.36%. We researched the impact of the comments on the status message only on the where the accuracy turned to be 67.31%. Adding the relationships of the friends i.e. close friends, family and general friends who made the comments to our experiments the accuracy turned to be 72.035 %. Finally, we researched the impact of the likes of the status messages taking into consideration the relationships of the friends who made the likes and comments of the status message taking into consideration the relationships of the friends who made the comments on the emotions of the social networks, We found out that the accuracy turned to be 75.23 %.
Computer Science & Engineering Department
MS in Computer Science
Committee Member 1
Committee Member 2
Library of Congress Subject Heading 1
Library of Congress Subject Heading 2
Emotions -- Mathematical models.
The author retains all rights with regard to copyright. The author certifies that written permission from the owner(s) of third-party copyrighted matter included in the thesis, dissertation, paper, or record of study has been obtained. The author further certifies that IRB approval has been obtained for this thesis, or that IRB approval is not necessary for this thesis. Insofar as this thesis, dissertation, paper, or record of study is an educational record as defined in the Family Educational Rights and Privacy Act (FERPA) (20 USC 1232g), the author has granted consent to disclosure of it to anyone who requests a copy.
Institutional Review Board (IRB) Approval
Approval has been obtained for this item
(2015).Enhancing Emotion Elicitation using the Contextual, Multimodal Features of a Social Network [Master's Thesis, the American University in Cairo]. AUC Knowledge Fountain.
Rizk, Ahmed. Enhancing Emotion Elicitation using the Contextual, Multimodal Features of a Social Network. 2015. American University in Cairo, Master's Thesis. AUC Knowledge Fountain.