| Autor: |
Zargari, Hamed, Hosseini, Mohammad Mehdi, Gharahbagh, Abdorreza Alavi |
| Předmět: |
|
| Zdroj: |
Multimedia Tools & Applications; May2024, Vol. 83 Issue 18, p54885-54907, 23p |
| Abstrakt: |
There are many natural language processing methods that can be used to analyze audio, image, and video captions and improve the accuracy of multimedia understanding tools. Linguistic phenomena are considered as one of the most effective factors in changing the initial score of sentiment words. Lacking accurate attention to the impact of intensifiers and their position in a sentiment component and neglecting the mutual effect between the polarity of sentiment words and intensifiers reduce the efficiency of sentiment dictionaries. To solve these problems, in the proposed method, a new high coverage semi-supervised sentiment dictionary was made from a combination of sentimental words and several intensifiers. The dictionary has several advantages, including assigning four fuzzy coefficients to intensifiers based on their position in sentiment compounds and the polarity of sentiment words. Sentimental word scores were extracted from the manual VADER dictionary, and the intensifiers coefficients were extracted automatically using a hybrid fuzzy-statistical method based on the intensifier location and the sentiment word polarity. The simulation results show the proposed dictionary has at least 9% improvement compared to the latest sentiment dictionaries at standard databases. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
Full text from SpringerLink