Date of Award


Publication Type

Master Thesis

Degree Name



Computer Science

First Advisor

Ezeife, Christie


Aspect-based Opinion Mining, Data Mining, Machine Learning, Microblogs, Opinion Mining, Text Mining




Aspect-based Opinion Mining (ABOM) systems take as input a corpus about a product and aim to mine the aspects (the features or parts) of the product and obtain the opinions of each aspect (how positive or negative the appraisal or emotions towards the aspect is). A few systems like Twitter Aspect Classifier and Twitter Summarization Framework have been proposed to perform ABOM on microblogs. However, the accuracy of these techniques are easily affected by spam posts and buzzwords. In this thesis we address this problem of removing noisy aspects in ABOM by proposing an algorithm called Microblog Aspect Miner (MAM). MAM classifies the microblog posts into subjective and objective posts, represents the frequent nouns in the subjective posts as vectors, and then clusters them to obtain relevant aspects of the product. MAM achieves a 50% improvement in accuracy in obtaining relevant aspects of products compared to previous systems.