University of Twente Student Theses


Influence of neutral word removal on sentiment analysis

Rietvelt, D.C.J.C. (2019) Influence of neutral word removal on sentiment analysis.

[img] PDF
Abstract:Nowadays, lots of information is shared on the Internet, including opinions. In order to deal with these opinions, computers perform sentiment analysis, which provides insight into the sentiment of a piece of text. Currently, not a lot of research is done for Dutch sentiment analysis. Preliminary experiments found that a larger lexicon enhances sentiment analysis, but this increase was too small compared to previous literature. A possible cause for this is the large amount of neutral words the larger lexicon contained after extension. This paper will investigate the influence of neutral word removal on the performance of sentiment analysis. Two experiments were conducted, one on an unbalanced dataset and one on a balanced dataset. Neutral words were gradually removed from the extended lexicon and the performance was measured. The two experiments both show that neutral word removal enhances sentiment analysis, but the differences are small. Furthermore, the experiment on the balanced dataset shows that a larger lexicon does not enhance sentiment analysis. Due to small enhancements and the opposite results compared to literature, no conclusion is drawn.
Item Type:Essay (Bachelor)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Business & IT BSc (56066)
Keywords:Sentiment analysis, Lexicon-based, Neutral word removal
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page