Mining opinions from reviews has been a field of ever-growing research. These include mining opinions on document level, sentence level and even aspect level. While explicitly mentioned aspects from user-generated texts have been widely researched, very little work has been done in gathering opinions on aspects that are implied and not explicitly mentioned. Previous work to identify implicit aspects and opinion was limited to syntactic-based classifiers or other machine learning methods trained on restaurant dataset. In this paper, the present is a novel study for extracting and analysing implicit aspects and opinions from airline reviews in English. Through this study, an airline domain-specific aspect-based annotated corpus, and a novel two-way technique that first augments pre-trained word embeddings for sequential with stochastic gradient descent optimized conditional random fields (CRF) and second using machine and ensemble learning algorithms to classify the implied aspects is devised and developed. This two-way technique resolves double-implicit problem, most encountered by previous work in implicit aspect and opinion text mining. Experiments with a hold-out test set on the first level i.e., entity extraction by optimized CRF yield a result of ROC-AUC score of 96% and F1 score of 94% outperforming few baseline systems. Further experiments with a range of machine and ensemble learning classifier algorithms to classify implied aspects and opinions for each entity yields a result of ROC-AUC score ranging from 71 to 94.8% for all implied entities. This two-level technique for implicit aspect extraction and classification outperforms many baseline systems in this domain.

Cyberbullying is the wilful and repeated infliction of harm on an individual using the Internet and digital technologies. Similar to face-to-face bullying, cyberbullying can be captured formally using the Routine Activities Model (RAM) whereby the potential victim and bully are brought into proximity of one another via the interaction on online social networking (OSN) platforms. Although the impact of the COVID-19 (SARS-CoV-2) restrictions on the online presence of minors has yet to be fully grasped, studies have reported that 44% of pre-adolescents have encountered more cyberbullying incidents during the COVID-19 lockdown. Transparency reports shared by OSN companies indicate an increased take-downs of cyberbullying-related comments, posts or content by artificially intelligen moderation tools. However, in order to efficiently and effectively detect or identify whether a social media post or comment qualifies as cyberbullying, there are a number factors based on the RAM, which must be taken into account, which includes the identification of cyberbullying roles and forms. This demands the acquisition of large amounts of fine-grained annotated data which is costly and ethically challenging to produce. In addition where fine-grained datasets do exist they may be unavailable in the target language. Manual translation is costly and expensive, however, state-of-the-art neural machine translation offers a workaround. This study presents a first of its kind experiment in leveraging machine translation to automatically translate a unique pre-adolescent cyberbullying gold standard dataset in Italian with fine-grained annotations into English for training and testing a native binary classifier for pre-adolescent cyberbullying. In addition to contributing high-quality English reference translation of the source gold standard, our experiments indicate that the performance of our target binary classifier when trained on machine-translated English output is on par with the source (Italian) classifier.

Cyberbullying remains a significant problem for children that appears to have been exacerbated by Covid-19 related lockdowns, which moved a lot of children's offline activities online. Transparency reports shared by social network and gaming platform companies indicate increased take-downs of offensive and harmful comments, posts or content by artificially intelligent (AI) tools. Nonetheless, little is known about how such tools are designed and developed, what data they are trained on; and how this is done in practice. Many studies have discussed the opacity of such algorithmic _moderation, detection and prevention_ solutions, and called for their greater transparency to understand *how* and *what* user-interactive engagement features help in AI decision-making. This study examines a) the use of AI-solutions by social media and gaming companies to proactively address cyberbullying on their platforms, b) explore the current cyberbullying detection, prevention and proactive intervention strategies by such companies, and c) through comprehensive database search, review of existing computational literature on monitoring, detection and intervention strategies to address cyberbullying incidents amongst children. Our findings show that *very scarce* resources are available in the public domain to build AI algorithmic solutions to combat cyberbullying, and _little information_ is publicly available that would allow scrutiny of platforms' enforcement mechanisms.

No abstract

This paper presents multidimensional Social Opinion Mining on user-generated content gathered from newswires and social networking services in three different languages: English -a high-resourced language, Maltese -a low-resourced language, and Maltese-English -a code-switched language. Multiple fine-tuned neural classification language models which cater for the i) English, Maltese and Maltese-English languages as well as ii) five different social opinion dimensions, namely subjectivity, sentiment polarity, emotion, irony and sarcasm, are presented. Results per classification model for each social opinion dimension are discussed.