Towards the Discovery of Influencers to Follow in Micro-Blogs (Twitter) by Detecting Topics in Posted Messages (Tweets)
Micro-blogs, such as Twitter, have become important tools to share opinions and information among users. Messages concerning any topic are daily posted. A message posted by a given user reaches all the users that decided to follow her/him. Some users post many messages, because they aim at being rec...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2020-08-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/10/16/5715 |
| _version_ | 1797557398112567296 |
|---|---|
| author | Mubashir Ali Anees Baqir Giuseppe Psaila Sayyam Malik |
| author_facet | Mubashir Ali Anees Baqir Giuseppe Psaila Sayyam Malik |
| author_sort | Mubashir Ali |
| collection | DOAJ |
| description | Micro-blogs, such as Twitter, have become important tools to share opinions and information among users. Messages concerning any topic are daily posted. A message posted by a given user reaches all the users that decided to follow her/him. Some users post many messages, because they aim at being recognized as influencers, typically on specific topics. How a user can discover influencers concerned with her/his interest? Micro-blog apps and web sites lack a functionality to recommend users with influencers, on the basis of the content of posted messages. In this paper, we envision such a scenario and we identify the problem that constitutes the basic brick for developing a recommender of (possibly influencer) users: training a classification model by exploiting messages labeled with topical classes, so as this model can be used to classify unlabeled messages, to let the hidden topic they talk about emerge. Specifically, the paper reports the investigation activity we performed to demonstrate the suitability of our idea. To perform the investigation, we developed an investigation framework that exploits various patterns for extracting features from within messages (labeled with topical classes) in conjunction with the mostly-used classifiers for text classification problems. By means of the investigation framework, we were able to perform a large pool of experiments, that allowed us to evaluate all the combinations of feature patterns with classifiers. By means of a cost-benefit function called “Suitability”, that combines accuracy with execution time, we were able to demonstrate that a technique for discovering topics from within messages suitable for the application context is available. |
| first_indexed | 2024-03-10T17:16:30Z |
| format | Article |
| id | doaj.art-a9bcc5fefea7426e931acd796e3d6053 |
| institution | Directory Open Access Journal |
| issn | 2076-3417 |
| language | English |
| last_indexed | 2024-03-10T17:16:30Z |
| publishDate | 2020-08-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj.art-a9bcc5fefea7426e931acd796e3d60532023-11-20T10:30:02ZengMDPI AGApplied Sciences2076-34172020-08-011016571510.3390/app10165715Towards the Discovery of Influencers to Follow in Micro-Blogs (Twitter) by Detecting Topics in Posted Messages (Tweets)Mubashir Ali0Anees Baqir1Giuseppe Psaila2Sayyam Malik3Department of Management, Information and Production Engineering, University of Bergamo, 24129 Bergamo, ItalyFaculty of Computing & IT, University of Sialkot, Sialkot 51040, PakistanDepartment of Management, Information and Production Engineering, University of Bergamo, 24129 Bergamo, ItalyFaculty of Computing & IT, University of Sialkot, Sialkot 51040, PakistanMicro-blogs, such as Twitter, have become important tools to share opinions and information among users. Messages concerning any topic are daily posted. A message posted by a given user reaches all the users that decided to follow her/him. Some users post many messages, because they aim at being recognized as influencers, typically on specific topics. How a user can discover influencers concerned with her/his interest? Micro-blog apps and web sites lack a functionality to recommend users with influencers, on the basis of the content of posted messages. In this paper, we envision such a scenario and we identify the problem that constitutes the basic brick for developing a recommender of (possibly influencer) users: training a classification model by exploiting messages labeled with topical classes, so as this model can be used to classify unlabeled messages, to let the hidden topic they talk about emerge. Specifically, the paper reports the investigation activity we performed to demonstrate the suitability of our idea. To perform the investigation, we developed an investigation framework that exploits various patterns for extracting features from within messages (labeled with topical classes) in conjunction with the mostly-used classifiers for text classification problems. By means of the investigation framework, we were able to perform a large pool of experiments, that allowed us to evaluate all the combinations of feature patterns with classifiers. By means of a cost-benefit function called “Suitability”, that combines accuracy with execution time, we were able to demonstrate that a technique for discovering topics from within messages suitable for the application context is available.https://www.mdpi.com/2076-3417/10/16/5715social mediamicro-blogs (Twitter)towards recommending influencers based on topic classificationinvestigation frameworkcomparison of various techniques for topic classificationcost-benefit function |
| spellingShingle | Mubashir Ali Anees Baqir Giuseppe Psaila Sayyam Malik Towards the Discovery of Influencers to Follow in Micro-Blogs (Twitter) by Detecting Topics in Posted Messages (Tweets) Applied Sciences social media micro-blogs (Twitter) towards recommending influencers based on topic classification investigation framework comparison of various techniques for topic classification cost-benefit function |
| title | Towards the Discovery of Influencers to Follow in Micro-Blogs (Twitter) by Detecting Topics in Posted Messages (Tweets) |
| title_full | Towards the Discovery of Influencers to Follow in Micro-Blogs (Twitter) by Detecting Topics in Posted Messages (Tweets) |
| title_fullStr | Towards the Discovery of Influencers to Follow in Micro-Blogs (Twitter) by Detecting Topics in Posted Messages (Tweets) |
| title_full_unstemmed | Towards the Discovery of Influencers to Follow in Micro-Blogs (Twitter) by Detecting Topics in Posted Messages (Tweets) |
| title_short | Towards the Discovery of Influencers to Follow in Micro-Blogs (Twitter) by Detecting Topics in Posted Messages (Tweets) |
| title_sort | towards the discovery of influencers to follow in micro blogs twitter by detecting topics in posted messages tweets |
| topic | social media micro-blogs (Twitter) towards recommending influencers based on topic classification investigation framework comparison of various techniques for topic classification cost-benefit function |
| url | https://www.mdpi.com/2076-3417/10/16/5715 |
| work_keys_str_mv | AT mubashirali towardsthediscoveryofinfluencerstofollowinmicroblogstwitterbydetectingtopicsinpostedmessagestweets AT aneesbaqir towardsthediscoveryofinfluencerstofollowinmicroblogstwitterbydetectingtopicsinpostedmessagestweets AT giuseppepsaila towardsthediscoveryofinfluencerstofollowinmicroblogstwitterbydetectingtopicsinpostedmessagestweets AT sayyammalik towardsthediscoveryofinfluencerstofollowinmicroblogstwitterbydetectingtopicsinpostedmessagestweets |