Date of Award

Fall 2021

Publication Type


Degree Name



Computer Science


Data Scientist, Machine Learning, Python Development, Research and Development, Software Developer




P. Zadeh



Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.


Existing OM systems like CONE take a partial historical rating of users on multiple products and perform opinion estimation to maximizes overall positive opinions using OM. However, CONE does not consider actual user opinions from social posts where users provide opinions through comments, likes and sharing about a product. OBIN mines users' low-frequency features from comments to create a community preference influence network utilizing user response on posts and relationships between them. However, OBIN only performs feature-level opinion mining and does not consider a joint approach that combines sentence-level and feature-level to remove subjective reviews and includes slang words and emoticons, which users often use over the internet. Also. OBIN discovers community preference but does not perform OM, which is more profitable to the seller when introducing a product in the market. The limitation of CONE and OBIN is that they consider opinion mining and maximization as separate subtasks that require more training time and do not consider community opinion, influence among the community users, nor use opinion maximization on their network to minimize viral marketing budget for selecting most influential nodes.

This thesis proposes Active Community Opinion Network Mining and Maximization (ACOMax), an extension of the OBIN system that adds active OM and joint opinion mining for solving two tasks (feature and sentence opinion mining) to enhance the model's accuracy by reducing training time. ACOMax first performs mining of multiple posts related to the product using TwitterAPI while considering relationships between users. Second, opinion mining (positive and neutral) from user reviews on selected posts to perform (i) Sentence-level mining to determine the overall positive sentiment of subjective opinions using VADER. (ii) feature-level opinion mining to extract frequent features with a favourable opinion about the product using the Apriori algorithm. Third, construct an opinion network graph of users who share positive opinions from (ii) to be utilized by the seller to actively select top k seed users with maximum opinion spread under Multiple Linear Threshold (MLT) for opinion maximization. To evaluate our model's performance, we extracted real-time user data using TwitterAPI. Our proposed model (ACOMax) outperforms previous models for total opinion spread in terms of F1 and Accuracy with the help of joint opinion mining and solves the cold start problem of CONE, and improves the total opinion spread in a social network.