Startseite Mathematik Assessment of fake news detection from machine learning and deep learning techniques
Kapitel
Lizenziert
Nicht lizenziert Erfordert eine Authentifizierung

Assessment of fake news detection from machine learning and deep learning techniques

  • Megha Shah , Akshay Kumar , Shristi Kharola und Mangey Ram
Veröffentlichen auch Sie bei De Gruyter Brill
Algorithms
Ein Kapitel aus dem Buch Algorithms

Abstract

The overwhelming prevalence of news data across numerous online platforms is far less taxing, but it has become an unnecessary burden that has transformed people’s lives. The dominance of mass media on the internet has had a substantial impact, with many individuals depending on it as a regular source of information. However, this widespread influence has also given rise to the creation of partially evident or entirely fabricated news stories. Intentionally spreading these fake stories through online social networking sites has become a common practice. Websites now primarily aim to mold public opinion using false information. The core aim of this study is to develop a reliable model that can identify a given news report as true or false. For this, the authors have modeled a trust-based architecture for online shared news incorporating natural language processing in machine learning (ML) and deep learning (DL) techniques. To develop the architecture, six ML models, two long-short-term memory (LSTM) models, and two distinct feature extraction techniques have been utilized. The findings reveal that, out of all six ML models, random forest with TF-IDF and logistic regression with CountVectorizer yield the optimal results. In the case of DL models, the outcomes for the LSTM model and bi-directional LSTM have yielded the same results.

Abstract

The overwhelming prevalence of news data across numerous online platforms is far less taxing, but it has become an unnecessary burden that has transformed people’s lives. The dominance of mass media on the internet has had a substantial impact, with many individuals depending on it as a regular source of information. However, this widespread influence has also given rise to the creation of partially evident or entirely fabricated news stories. Intentionally spreading these fake stories through online social networking sites has become a common practice. Websites now primarily aim to mold public opinion using false information. The core aim of this study is to develop a reliable model that can identify a given news report as true or false. For this, the authors have modeled a trust-based architecture for online shared news incorporating natural language processing in machine learning (ML) and deep learning (DL) techniques. To develop the architecture, six ML models, two long-short-term memory (LSTM) models, and two distinct feature extraction techniques have been utilized. The findings reveal that, out of all six ML models, random forest with TF-IDF and logistic regression with CountVectorizer yield the optimal results. In the case of DL models, the outcomes for the LSTM model and bi-directional LSTM have yielded the same results.

Heruntergeladen am 20.10.2025 von https://www.degruyterbrill.com/document/doi/10.1515/9783111229157-006/html?lang=de
Button zum nach oben scrollen