Analysing Emotional Sentiment in People s YouTube Channel Comments

Transcription

1 Analysing Emotional Sentiment in People s YouTube Channel Comments Eleanor Mulholland, Paul Mc Kevitt, and Tom Lunney Ulster University School of Creative Arts & Technologies Derry/Londonderry, Northern Ireland Karl-Michael Schneider Google Ireland Ltd. Barrow Street, Dublin, Ireland

2 Background & Literature Review 360-MAM-Select 360-MAM-Affect Results Introduction YouTube Channel video URLs harvested Analysing Emotional Sentiment Discussion Relation to Other Work Conclusion & Future Work Acknowledgements

3 Recommender Systems Rank products against others for comparison Faridani (2011) trained a recommender model with ratings from OpinionSpace dataset Hanser et al. (2010) implemented NewsViz giving numerical emotion ratings to words Tkalčič et al. (2011) propose a Unifying Framework for emotion detection

4 Unifiying Framework (Tkalčič et al. 2011)!

5 Sentiment Analysis Recognising negative, positive & neutral opinions (Wilson et al. 2005) Methods of opinion collecting (Khan 2009) Variety of data on the Internet in different forms (Nasukawa & Yi 2003) Online information is not static (Khan et al. 2009) NLP and Machine Learning (Tzanis et al. 2006)

6 Emotions Experience of mood has two elements (Minsky 2007) Meta-emotions as true reflection (Willaert et al. 2013) Control-Value Theory & emotions (Muñoz et al. 2016) Lerner et al. (2015) proposed the Emotion-Imbued Choice Model with integral emotions & incidental emotions

7 Emotion-Imbued Choice Model (Lerner et al. 2015)

8 Gamification Game mechanics & game design techniques used to enhance non-game scenarios (Deterding et al. 2011) Gamification is popular for monitoring & analysing online communities (Bista et al. 2012) Gamification has improved learning & information retention in education (Landers & Callan 2011 Charles et al. 2011) Chou (2015) created the Octalysis gamification design framework

9 Octalysis Gamification Design Framework (Chou 2015)

10 360-MAM-Select How user responds emotionally to media content Modules for sentiment analysis & emotion modelling (360-MAM-Affect) & gamification (360-Gamify) 360-MAM-Affect harvests YouTube comments on video content Identifying the overall reception of a video, providing tailored recommendations for particular users

11 Architecture of 360-MAM-Select

12 360-MAM-Affect YouTube comments harvested with YouTube API (Google 2015) for YouTube data GATE (Cunningham 2015) for natural language processing EmoSenticNet (Gelbukh 2014) for identifying emotion words RapidMiner (RapidMiner 2015) for counting the average frequency of emotion words identified

13 YouTube Channel video URLs harvested

14 Average emotion frequencies across YouTube channels

15 Discussion 3 (Sad, Surprise & Joy) of the 7 emotion tags had average emotion frequency > 20 Anger, Disgust, Fear & Neutral had considerably lower frequencies with none of these 4 emotions having average emotion frequency > 3 Neutral, with average emotion frequency of only 1 is not unexpected due to low number of neutral concepts in EmoSenticNet

16 Relation to Other Work Recommender systems (Ricci et al. 2011) ability to personalise experiences (Linden et al. 2003) to provide high quality recommendations (Bartłomiej 2005) Emotion has been identified as important factor in improving recommender systems (Tkalčic et al. 2011) Polignano (2015) proposes a framework to include emotions inside recommender systems 360-MAM-Select will advance recommender systems by providing improved user experience

17 Conclusion & Future Work Sentiment analysis, emotion detection & modelling & gamification will improve online recommendation of media assets 360-MAM-Affect successfully tagged YouTube comments with EmoSenticNet to identify sentiment with Anger, Disgust, Joy, Sad, Surprise, Fear & Neutral Further implementation & testing of 360-MAM-Select employing the Unifying Framework (Tkalčič et al. 2011) & Emotion-Imbued Choice (EIC) model (Lerner et al. 2015) within 360-MAM-Affect 360-Gamify will be implemented & tested with the Octalysis gamification framework (Chou 2015)

18 Acknowledgements Dr. Brian Bridges, Ulster Univeristy Dr. Kevin Curran, Ulster Univeristy Dr. Lisa Fitzpatrick, Ulster Univeristy Dr. Karla Muñoz-Esquivel, BijouTech 360 Production Ltd. & Alleycats TV Northern Ireland Department of Employment & Learning (DEL) Co-operative Awards in Science & Technology (CAST) Ph.D. Award