Building a subjectivity lexicon for Filipino
Textual information all domain can be categorized into two namely, facts (objective information) and opinion (subjective information). Facts contain objective information about an entity as well as its attributes and properties. Opinions, on the other hand, are subjective information which contains...
Saved in:
| Main Author: | |
|---|---|
| Format: | text |
| Language: | English |
| Published: |
Animo Repository
2013
|
| Subjects: | |
| Online Access: | https://animorepository.dlsu.edu.ph/etd_masteral/6824 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | De La Salle University |
| Language: | English |
| Summary: | Textual information all domain can be categorized into two namely, facts (objective information) and opinion (subjective information). Facts contain objective information about an entity as well as its attributes and properties. Opinions, on the other hand, are subjective information which contains description of people’s emotion towards an entity as well as its attributes and properties. Though there are existing lexicons that would assist sentiment analysis, this lexicons are not solely built to support analysis for the Filipino language. Thus this research was aimed to address the problem of limited resources for sentiment analysis in Filipino by building a subjectivity lexicon extracted from Filipino opinion articles. A machine-learning based approach for lexicon expansion was adapted in this research. A classifier was built through Weka using three different machine learning algorithms, namely, C4.5 (decision tree), Naïve Bayes, and k-Nearest Neighbor; and was evaluated using 10-cross fold validation. Results show that k- Nearest Neighbor gave the best result, with 92.20% and 99.42% for window 5 and window 3 respectively. |
|---|