| Autor: |
Bhaskar, Vijay, Shanmugam, Udhayakumar |
| Předmět: |
|
| Zdroj: |
AIP Conference Proceedings; 2023, Vol. 2655 Issue 1, p1-9, 9p |
| Abstrakt: |
To detect spam comments from normal comments in YouTube using novel count vectorizer techniques with SVM and to improve the recall and precision values in comparison with Naive Bayes filtering techniques for YouTube channel popularity. Two sample groups are taken into consideration and tested, G-power is considered as 80 for calculating the sample size and for t-test analysis. Overall 1956 comments in each sample are tested for spams. The feature extraction using count vectoriser based SVM has an average accuracy of 96%, which seems to appear better recall and precision values over traditional Naive Bayes method. Since the significance is around 0.218, there is a statistically significant difference among the study group with (p<0.05). The increase in precision value of detecting the spam can substantially make the youtube viewership more popular and help prevent irrelevant content. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
