English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

With the growth of social medias, such as Twitter, plenty of user-generateddata emerge daily. The short texts published on Twitter -- the tweets -- haveearned significant attention as a rich source of information to guide manydecision-making processes. However, their inherent characteristics, such as theinformal, and noisy linguistic style, remain challenging to many naturallanguage processing (NLP) tasks, including sentiment analysis. Sentimentclassification is tackled mainly by machine learning-based classifiers. Theliterature has adopted word representations from distinct natures to transformtweets to vector-based inputs to feed sentiment classifiers. Therepresentations come from simple count-based methods, such as bag-of-words, tomore sophisticated ones, such as BERTweet, built upon the trendy BERTarchitecture. Nevertheless, most studies mainly focus on evaluating thosemodels using only a small number of datasets. Despite the progress made inrecent years in language modelling, there is still a gap regarding a robustevaluation of induced embeddings applied to sentiment analysis on tweets.Furthermore, while fine-tuning the model from downstream tasks is prominentnowadays, less attention has been given to adjustments based on the specificlinguistic style of the data. In this context, this study fulfils an assessmentof existing language models in distinguishing the sentiment expressed in tweetsby using a rich collection of 22 datasets from distinct domains and fiveclassification algorithms. The evaluation includes static and contextualizedrepresentations. Contexts are assembled from Transformer-based autoencodermodels that are also fine-tuned based on the masked language model task, usinga plethora of strategies.

Sentiment analysis in tweets: an assessment study from classical to modern text representation models