Sanitization of septic news sentences through hybrid approach in English.
Saved in:
| Title: | Sanitization of septic news sentences through hybrid approach in English. |
|---|---|
| Authors: | Das, Soma1,2 (AUTHOR) soma_phd_2018july@iiitkalyani.ac.in, Chatterji, Sanjay1 (AUTHOR) sanjayc@iiitkalyani.ac.in |
| Source: | Language Resources & Evaluation. Sep2025, Vol. 59 Issue 3, p1865-1897. 33p. |
| Subjects: | Machine learning, Paraphrase, Text processing (Computer science), Fake news, Information filtering |
| Abstract: | News articles play an important role in shaping public opinion and influencing decision-making. Sentences of standard news articles are often manipulated to favour a person, group, or political party or reflect a particular sentiment or agenda. It is challenging to define and filter or sanitize such news content before presenting it to readers. In our research, we focus on addressing some of the important issues of problematic English news sentences referred to as Septic sentences. With the aid of Machine Learning algorithms, we have successfully identified these sentences and their corresponding Septic phrases. We sanitize these Septic sentences by converting them into Pure sentences. In our paper, we demonstrate the sanitization process using a hybrid system, i.e., a rule-based approach followed by paraphrasing techniques. We evaluate our models using both syntactic and semantic similarity measured. We leverage the GPT - 3.5 and Parrot models in the direct paraphrasing approach. For Indirect paraphrasing, we use Google Translation API to translate the Septic English sentences into Spanish, German, and Tagalog, followed by back translation into English sentences. Additionally, we use the DeepL API to perform the same task through Spanish. Overall, DeepL model gives the highest accuracy throughout the metrics compared to the other Direct and Indirect methods. [ABSTRACT FROM AUTHOR] |
| Copyright of Language Resources & Evaluation is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | News articles play an important role in shaping public opinion and influencing decision-making. Sentences of standard news articles are often manipulated to favour a person, group, or political party or reflect a particular sentiment or agenda. It is challenging to define and filter or sanitize such news content before presenting it to readers. In our research, we focus on addressing some of the important issues of problematic English news sentences referred to as Septic sentences. With the aid of Machine Learning algorithms, we have successfully identified these sentences and their corresponding Septic phrases. We sanitize these Septic sentences by converting them into Pure sentences. In our paper, we demonstrate the sanitization process using a hybrid system, i.e., a rule-based approach followed by paraphrasing techniques. We evaluate our models using both syntactic and semantic similarity measured. We leverage the GPT - 3.5 and Parrot models in the direct paraphrasing approach. For Indirect paraphrasing, we use Google Translation API to translate the Septic English sentences into Spanish, German, and Tagalog, followed by back translation into English sentences. Additionally, we use the DeepL API to perform the same task through Spanish. Overall, DeepL model gives the highest accuracy throughout the metrics compared to the other Direct and Indirect methods. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 1574020X |
| DOI: | 10.1007/s10579-024-09778-0 |
