A Framework to Predict the Quality of a Video for Popularity on Social Media
ABSTRACT YouTube has become a dominant force in digital media, yet current video popularity analytics remain limited in capturing the emotional and cultural dimensions of viewer engagement, particularly in underrepresented regions like Pakistan. While existing research focuses predominantly on Weste...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2025-06-01
|
| Series: | Engineering Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1002/eng2.70250 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849417619506987008 |
|---|---|
| author | Abqa Javed Nimra Abid Muhammad Shoaib Muhammad Farrukh Shahzad Fahad Sabah Raheem Sarwar |
| author_facet | Abqa Javed Nimra Abid Muhammad Shoaib Muhammad Farrukh Shahzad Fahad Sabah Raheem Sarwar |
| author_sort | Abqa Javed |
| collection | DOAJ |
| description | ABSTRACT YouTube has become a dominant force in digital media, yet current video popularity analytics remain limited in capturing the emotional and cultural dimensions of viewer engagement, particularly in underrepresented regions like Pakistan. While existing research focuses predominantly on Western markets and quantitative metrics (views, likes, comments), these approaches overlook sentiment‐driven interactions critical to understanding regional audience behavior. This study bridges this gap by introducing a sentiment‐aware framework for YouTube video classification in Pakistan, combining traditional popularity metrics with advanced sentiment analysis of user comments. We curated the PAK VIDEOS (2021–2023) dataset using YouTube Data APIs, comprising metadata and user comments from Pakistan's top trending videos. Leveraging Natural Language Processing (NLP) techniques, we extracted sentiment scores from comments to classify videos into four categories: non‐popular, overwhelmingly positive, overwhelmingly negative, and neutral. This hybrid approach enabled a nuanced evaluation of content reception beyond quantitative metrics. Four machine learning models—random forest, stochastic gradient descent classifier (SGDC), gradient boosting, and XGBoost—were evaluated for classification. XGBoost achieved superior performance (84.3% accuracy), outperforming baseline models by up to 20%. Our framework demonstrates that integrating sentiment analysis significantly enhances popularity prediction, particularly in culturally distinct contexts. |
| format | Article |
| id | doaj-art-438ede7ad18245c6a5ec7ebb9ee2be5e |
| institution | Kabale University |
| issn | 2577-8196 |
| language | English |
| publishDate | 2025-06-01 |
| publisher | Wiley |
| record_format | Article |
| series | Engineering Reports |
| spelling | doaj-art-438ede7ad18245c6a5ec7ebb9ee2be5e2025-08-20T03:32:45ZengWileyEngineering Reports2577-81962025-06-0176n/an/a10.1002/eng2.70250A Framework to Predict the Quality of a Video for Popularity on Social MediaAbqa Javed0Nimra Abid1Muhammad Shoaib2Muhammad Farrukh Shahzad3Fahad Sabah4Raheem Sarwar5Department of Computer Science University of Engineering and Technology Lahore PakistanDepartment of Computer Science University of Engineering and Technology Lahore PakistanDepartment of Computer Science University of Engineering and Technology Lahore PakistanCollege of Economics & Management Beijing University of Technology Beijing ChinaCollege of Computer Science Beijing University of Technology Beijing ChinaOTEHM, Faculty of Business and Law Manchester Metropolitan University Manchester UKABSTRACT YouTube has become a dominant force in digital media, yet current video popularity analytics remain limited in capturing the emotional and cultural dimensions of viewer engagement, particularly in underrepresented regions like Pakistan. While existing research focuses predominantly on Western markets and quantitative metrics (views, likes, comments), these approaches overlook sentiment‐driven interactions critical to understanding regional audience behavior. This study bridges this gap by introducing a sentiment‐aware framework for YouTube video classification in Pakistan, combining traditional popularity metrics with advanced sentiment analysis of user comments. We curated the PAK VIDEOS (2021–2023) dataset using YouTube Data APIs, comprising metadata and user comments from Pakistan's top trending videos. Leveraging Natural Language Processing (NLP) techniques, we extracted sentiment scores from comments to classify videos into four categories: non‐popular, overwhelmingly positive, overwhelmingly negative, and neutral. This hybrid approach enabled a nuanced evaluation of content reception beyond quantitative metrics. Four machine learning models—random forest, stochastic gradient descent classifier (SGDC), gradient boosting, and XGBoost—were evaluated for classification. XGBoost achieved superior performance (84.3% accuracy), outperforming baseline models by up to 20%. Our framework demonstrates that integrating sentiment analysis significantly enhances popularity prediction, particularly in culturally distinct contexts.https://doi.org/10.1002/eng2.70250Google Apps ScriptNatural Language Processingopinion miningsocial media analysisvideo classificationYouTube data APIs |
| spellingShingle | Abqa Javed Nimra Abid Muhammad Shoaib Muhammad Farrukh Shahzad Fahad Sabah Raheem Sarwar A Framework to Predict the Quality of a Video for Popularity on Social Media Engineering Reports Google Apps Script Natural Language Processing opinion mining social media analysis video classification YouTube data APIs |
| title | A Framework to Predict the Quality of a Video for Popularity on Social Media |
| title_full | A Framework to Predict the Quality of a Video for Popularity on Social Media |
| title_fullStr | A Framework to Predict the Quality of a Video for Popularity on Social Media |
| title_full_unstemmed | A Framework to Predict the Quality of a Video for Popularity on Social Media |
| title_short | A Framework to Predict the Quality of a Video for Popularity on Social Media |
| title_sort | framework to predict the quality of a video for popularity on social media |
| topic | Google Apps Script Natural Language Processing opinion mining social media analysis video classification YouTube data APIs |
| url | https://doi.org/10.1002/eng2.70250 |
| work_keys_str_mv | AT abqajaved aframeworktopredictthequalityofavideoforpopularityonsocialmedia AT nimraabid aframeworktopredictthequalityofavideoforpopularityonsocialmedia AT muhammadshoaib aframeworktopredictthequalityofavideoforpopularityonsocialmedia AT muhammadfarrukhshahzad aframeworktopredictthequalityofavideoforpopularityonsocialmedia AT fahadsabah aframeworktopredictthequalityofavideoforpopularityonsocialmedia AT raheemsarwar aframeworktopredictthequalityofavideoforpopularityonsocialmedia AT abqajaved frameworktopredictthequalityofavideoforpopularityonsocialmedia AT nimraabid frameworktopredictthequalityofavideoforpopularityonsocialmedia AT muhammadshoaib frameworktopredictthequalityofavideoforpopularityonsocialmedia AT muhammadfarrukhshahzad frameworktopredictthequalityofavideoforpopularityonsocialmedia AT fahadsabah frameworktopredictthequalityofavideoforpopularityonsocialmedia AT raheemsarwar frameworktopredictthequalityofavideoforpopularityonsocialmedia |