Please use this identifier to cite or link to this item: https://hdl.handle.net/10356/70158
Title: Anomaly detection through enhanced sentiment analysis on social media data
Authors: Zhao, Jingying
Keywords: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Document and text processing
Issue Date: 2017
Abstract: With the rising popularity of the use of social media, it is imperative for private companies and public organizations to analyze and understand the information posted by their users. One way of finding their attitude toward specific products or events is to use sentiment analysis. However, there are some abnormal behaviors embedded in the information available in social media, such as abnormal opinions and unusual patterns. One common type of anomaly is sarcasm. Sometimes feelings are expressed inexplicitly and sarcastically by users. Such sentiment of the sentences can be misled by words with strong polarity while the opposite meaning was intended. Therefore, the objective of this project is to identify a feasible method to detect sarcasm in social media, so that the sentiment analysis will be more accurate. A rule-based method was proposed in this project. By using lexicon-based sentiment analysis, patterns with the different sequence of sentiment polarity were discovered. The accuracy and other evaluation aspects were used to find out the capability of this method. The results have shown that this approach can reach a quite acceptable performance. Overall, the objective of this project has been successfully achieved, but there are still some limitations where improvement and further implementation will be needed in the future.
URI: http://hdl.handle.net/10356/70158
Rights: Nanyang Technological University
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:SCSE Student Reports (FYP/IA/PA/PI)

Files in This Item:
File Description SizeFormat 
FYP_Report_Zhao Jingying.pdf
  Restricted Access
912 kBAdobe PDFView/Open

Google ScholarTM

Check

Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.