Search

Irony is a complex linguistic phenomenon that has been extensively studied in computational linguistics across many languages. Existing research has relied heavily on annotated corpora, which are inherently biased due to their creation process. This study focuses on the problem of bias in cross-domain and cross-language irony detection and aims to identify the extent of topic bias in benchmark corpora and how it affects the generalization of models across domains and languages (English, Spanish, and Italian). Our findings offer a first insight into this issue and showed that mitigating the topic bias in these corpora improves the generalization of models beyond their training data. These results have important implications for the development of robust models in the analysis of ironic language.

Fake news are spread by exploiting specific linguistic patterns aimed at triggering negative emotions and persuading the consumers. A way to contrast this phenomenon is to analyse the psychological factors underlying consumers’ vulnerabilities. This paper is situated in this research context: first, we study the correlation between psycho-linguistic patterns in user’s posts and the tendency to spread false information. Moreover, since online contents exploit multimedia information, a methodology aimed at profiling the authors based on the images they share is employed. The reported experiments show that the proposed method, which considers both text-related and image-related features, outperforms the results of state-of-the-art approaches.

Hate speech may take different forms in online social media. Most of the investigations in the literature are focused on detecting abusive language in discussions about ethnicity, religion, gender identity and sexual orientation. In this paper, we address the problem of automatic detection and categorization of misogynous language in online social media. The main contribution of this paper is two-fold: (1) a corpus of misogynous tweets, labelled from different perspective and (2) an exploratory investigations on NLP features and ML models for detecting and classifying misogynistic language

In last years, the spread of social Web has promoted a strong interest in analyzing how information related to a given topic diffuses. Nevertheless, this is still quite an unexplored field in the literature. In this paper we propose a general approach that makes use of a set of Natural Language Processing (NLP) techniques to analyse some of the most important features of information related to a topic. The domain of this study is Twitter, since here topics are easily identified by means of hashtags. In particular, our aim is to analyse the possible change over time of the content sub-topicality and sentiment in the tracked tweets, and bring out their relationships with the users’ demographic features.

Information Extraction (IE) is a natural language processing (NLP) task whose aim is to analyze texts written in natural language to extract structured and useful information such as named entities and semantic relations linking these entities. Information extraction is an important task for many applications such as bio-medical literature mining, customer care, community websites, and personal information management. The increasing information available in patient clinical reports is difficult to access. As it is often in an unstructured text form, doctors need tools to enable them access to this information and the ability to search it. Hence, a system for extracting this information in a structured form can benefits healthcare professionals. The work presented in this paper uses a local grammar approach to extract medical named entities from French patient clinical reports. Experimental results show that the proposed approach achieved an F-Measure of 90. 06%.

One way to achieve semantic interoperability when data lexicalized in different languages is by means of cross-lingual linking. Translation resources are used as an intermediate step to reduce the language barriers. The key challenge is to select the correct mapping among candidate matches. We define an experiment to study upper bounds for the correctness of a cross-lingual ontology matching system. We highlight different lexical characteristics that can support the selection step. We believe that our findings can be useful in the design of cross-lingual mapping algorithms.

In this chapter they present a collaborative and situational tool called MUSTER, that has been specifically designed and developed for requirements elicitation workshops, and which utilizes, extends, and demonstrates a successful application of intelligent technologies for Computer Aided Software Engineering and Computer Aided Method Engineering. The primary objective of this tool is to improve the effectiveness and efficiency of the requirements elicitation process for software systems development, whilst addressing some of the common issues often encountered in practice through the integration of intelligent technologies. The tool also offers an example of how a group support system, coupled with artificial intelligence, can be applied to very practical activities and situations within the software development process.