UTB-NLP at SemEval-2023 Task 3: Weirdness, Lexical Features for Detecting Categorical Framings, and Persuasion in Online News

Proceedings of the The 17th International Workshop on Semantic Evaluation (SemEval-2023), pages 1551–1557 July 13-14, 2023 ©2023 Association for Computational Linguistics UTB-NLP at SemEval-2023 Task 3: Weirdness, Lexical Features for Detecting Categorical Framings, and Persuasion in Online News Juan Cuadrado and Elizabeth Martinez and Anderson Morillo and Daniel Pe˜na Kevin Sossa and Juan Carlos Martinez-Santos and Edwin Puertas Universidad Tenologica de Bolivar, Cartagena Colombia epuerta@utb.edu.co Abstract Nowadays, persuasive messages are increas- ingly frequent in social networks, which gener- ates particular concern in several communities, given that persuasion seeks to guide others towards adopting ideas, attitudes, or actions that they consider to be beneficial to them- selves. The efficient detection of news genre categories, detection of framing, and detection of persuasion techniques require several scien- tific disciplines, such as computational linguis- tics and sociology. Here we illustrate how we use lexical features given a news article to de- termine whether it is an opinion piece, aims to report factual news, or is satire. This paper presents a novel strategy for communication based on Lexical Weirdness. The results are part of our participation in Sub-Tasks 1 and 2 in SemEval 2023 Task 3. 1 Introduction Frame identification is a critical task in natural lan- guage processing (NLP) that recognizes the frames or semantic structures used in an article or text. In the SemEval 2023 challenge context, we propose the task of frame detection in news articles. In this case, we selected the English language. This task is essential because it allows a better understanding of the perspective and tone of news articles and other media. In addition, it can help detect bias and improve accuracy in news and opinion article clas- sification, which is relevant for applications such as news monitoring, sentiment analysis, and text classification. In particular, the frame detection challenge in which this paper seeks to promote the development of new NLP techniques and models for frame identification in complex and realistic texts and builds on the dataset and evaluation pre- sented in the challenge overview article. The approach presented in this paper is Lexical Weirdness. This approach seeks a more appropri- ate lexicon for the domain as we generated these lexicons focusing on having a more appropriate vocabulary. The first thing that we did was to cre- ate a corpus associated with each of the Framing categories. Then, we scrapped Wikipedia-related articles. Subsequently, we extracted the nominal uni- grams of these articles. After acquiring the uni- grams for each category, we calculated the rele- vance of words within the context by comparing the lexicon of each context with the lexicon gen- erated by Google. Finally, we selected the words with the highest occurrence, which had the most significant relevance in the context, to become the specific lexicon for each domain we implemented in a bag of words model. Finally, the frequency of words per category was obtained and combined with the frequencies of un- igrams using the Weirdness technique to generate a lexicon. From this lexicon constructed in the …

Content truncated to 500 words