Generic AI Framework to Identify and Analyze Influential Users in Online Social Networks Leveraging Dynamic Clustering Sentiment Analysis Temporal Influences and Deep Learning
| dc.contributor.guide | Bagavathi Sivakumar P and Shunmuga Velayutham C | |
| dc.coverage.spatial | ||
| dc.creator.researcher | Ramya G R | |
| dc.date.accessioned | 2025-12-17T11:50:42Z | |
| dc.date.available | 2025-12-17T11:50:42Z | |
| dc.date.awarded | 2025 | |
| dc.date.completed | 2025 | |
| dc.date.registered | 2014 | |
| dc.description.abstract | Identifying influential users (IUs) with substantial sway over others is essential for organizations aiming to leverage social media platforms for marketing and customer engagement. Traditionally, companies have established supporter networks to disseminate positive sentiments about their products and services. However, maximizing the spread of positive influence requires accurate identification of the most impactful users. Recent studies have increasingly focused on advancing IU detection techniques on platforms such as Twitter. The Temporal Influence Model (TIM) was introduced to evaluate evolving user opinions through complex interaction patterns, incorporating assumptions about resilient beliefs and influence decay over time. However, its reliance on computationally intensive Maximum Likelihood Estimation (MLE) renders it impractical for real-world applications. Additionally, the noise, redundancy,and sparsity inherent in social media data frequently result in inaccurate IU identification. This research seeks to overcome these limitations by enhancing topic detection, sentiment analysis, and influencer identification processes. Dynamic topic clustering and sentiment lexicons are employed to extract relevant themes and opinions from noisy Twitter data, while fuzzy neural networks and regression-based deep learning models automate feature extraction to improve prediction accuracy. Furthermore, analyzing user relationships and activity patterns within subgraphs provides additional signals for influencer identification. The proposed solutions refine TIM s capabilities, addressing the challenges posed by high-dimensional, unstructured social data. These advancements facilitate more precise and efficient IU detection, contributing significantly to the development of practical marketing tools and deeper insights into customer newlinebehavior on social platforms. Influential users are defined as individuals in a social network who significantly affect the opinions, behaviors, or actions of others through interactions.... | |
| dc.description.note | ||
| dc.format.accompanyingmaterial | None | |
| dc.format.dimensions | ||
| dc.format.extent | xx, 208 | |
| dc.identifier.researcherid | 0000-0001-8184-148X | |
| dc.identifier.uri | http://hdl.handle.net/10603/681763 | |
| dc.language | English | |
| dc.publisher.institution | Department of Computer Science and Engineering | |
| dc.publisher.place | Coimbatore | |
| dc.publisher.university | Amrita Vishwa Vidyapeetham University | |
| dc.relation | ||
| dc.rights | university | |
| dc.source.university | University | |
| dc.subject.keyword | Engineering and Technology:School of Computing; Computer Science and Engineering; | |
| dc.subject.keyword | Social Data; Sentiment analysis; Influential user; Weighted partition around medoids; Artificial cooperative search; Fuzzy deep neural network; Incremental learning logistic regression; Online social network; Temporal influential model; Weighted associated influence; User similarity matrix; User trustworthiness; Deep regression; Convolutional Neural Network; Social network; Deep Learning; | |
| dc.title | Generic AI Framework to Identify and Analyze Influential Users in Online Social Networks Leveraging Dynamic Clustering Sentiment Analysis Temporal Influences and Deep Learning | |
| dc.title.alternative | ||
| dc.type.degree | Ph.D. |
Files
Original bundle
1 - 5 of 15
Loading...
- Name:
- 01_title.pdf
- Size:
- 302.46 KB
- Format:
- Adobe Portable Document Format
- Description:
- Attached File
License bundle
1 - 1 of 1