Search

In recent years, social networks have changed the way people communicate, while the dissemination and obtaining of information has become fast and convenient. However, it is a double-edged sword, providing both a breeding ground for disinformation and a medium for its rapid dissemination. Meanwhile, falsehoods are proliferating and becoming a global risk, and deepfake technology makes it harder to discern. Therefore, this paper proposed a large language model-based algorithm to detect disinformation with generated videos and photos, which enhances the ability to discriminate fake news by implanting alignment and task specific instructions on the Llama model. Eventually, the result indicated that the proposed method has the potential to achieve superior performance and aligned with people's judgment of things. Furthermore, the paper delves into practical discussions concerning fine-tuning the model within the constraints of limited computational power, highlighting the challenges and potential solutions in optimizing the algorithm for applications. [ABSTRACT FROM AUTHOR]

Moving beyond classical single-label classification tasks proves to be crucial nowadays since the majority of real-life problems are inherently multi-label. Various tasks require involving, for one single instance as an input, several classes simultaneously or none of them. Particularly, detecting abusive texts in Arabic social networks intuitively carries several manifestations with regard to the targets' referred characteristics like gender, race, sexual orientation, etc. Existing multi-label algorithms either invest considerable complexity to handle labels' correlation or explicitly ignore this information to deal with each label as an independent binary problem. To remedy the abovementioned problems, this paper proposes a new multi-label classification solution based on deep neural networks. The main idea is to exploit the correlation between all the dataset's labels to break it into ordered subsets. Then, follow that order to implement a chain of BERT classifiers. Each classifier is therefore trained on a corresponding subset and passes its output via feature space to the following one. For a new instance, we gather all the classifiers' results and pick the best one for each class. Experiments showed that our approach significantly outperformed the single classifier's method using BERT as well as other neural network architectures on our Arabic abusive language dataset. 91%,76%, and 82% were achieved respectively for precision, recall, and F1-score. Moreover, we further investigated the performance of our approach on Kaggle multi-label research paper dataset and achieved very promising results. [ABSTRACT FROM AUTHOR]

With the fast growth of online content and social media platforms, the amount of user- generated data available in the form of product evaluations, social media postings, and online debates has rapidly expanded. It has become essential for organizations to comprehend client attitudes, make wise decisions, and improve their goods and services by analysing the sentiment indicated in this data. This paper aims to develop a sentiment analysis system for product reviews using URLs of particular item as data sources. It utilizes web scraping techniques to collect textual information from web pages. Natural Language Processing (NLP) techniques are applied for data pre-processing, including tokenization and stop word removal. Machine learning models like Support Vector Classifier (SVC), Random Forest, KNN, Logistic Regression, Naive Bayes and Deep learning models like LSTM and GRU are used for sentiment analysis. Performance measures such as recall, accuracy, and precision are employed to evaluate the models. High scores were consistently achieved by the LSTM and GRU models across all measures, demonstrating their potency in successfully capturing sequential relationships. With strong F-1 ratings of 91% and 90%, respectively, Long Short Term Memory (LSTM) and Gated Recurrent Units (GRU) regularly beat other models, demonstrating their outstanding precision-recall balance. Conversely, K-Nearest Neighbor (KNN) presents a significant trade-off, offering a blazing-fast training time of 0.01 seconds but demanding an extensive 102.34 seconds for prediction, resulting in a lower F-1 score of 47%. Support Vector Machine (SVM) and Logistic Regression achieve competitive F-1 scores of 85% and 84%, respectively, with relatively slow training and prediction times, striking a balance between performance and efficiency. Naive Bayes and Random Forest, while respectable in accuracy, fall slightly behind with F-1 scores of 71% and 80%, respectively, highlighting the significance of taking into account both performance and computing efficiency when choosing the best approach. Through this research, valuable insights are gained into the effectiveness and accuracy of different algorithms in sentiment analysis. The findings contribute to the understanding of sentiment analysis techniques and offer guidance for selecting the most suitable algorithm for URL-based sentiment analysis applications. [ABSTRACT FROM AUTHOR]

The outbreak of conflict between Russia and Ukraine on February 23, 2022, was a huge shock, and it quickly became the dominant topic of conversation on social media. Our research is based on Twitter conversations about the conflict between Russia and Ukraine. This paper intends to monitor and analyze the Twitter conversation on the continuing war in Ukraine using natural language processing techniques such as word cloud analysis, sentiment analysis, and visualization. The major goal is to acquire insight into Twitter users' shifting thoughts and opinions on various parts of the conflict over time. The study's findings have important implications for politicians, journalists, and social media users who want to follow and understand how online communities' attitudes on global issues change. We hope that this study will help to gain a deeper knowledge of how social media works in affecting public debate and how it might be used to gain insights into public opinion on crucial global issues. In this work logistic regression, K-nearest neighbours and decision tree algorithm are used to analyse the sentiments and got classification accuracy of 94.58%, 88.89% and 90.45% respectively. [ABSTRACT FROM AUTHOR]

Indonesian cuisine, especially Bakso, is one of the trendy foods that many travelers try to consume. Among the variations of Bakso, Bakso Malang is one of the most popular variants that is much talked about in social media. The present study analyzes the sentiment text mining of Bakso Malang business on post covid period from a natural language processing (NLP) social perspective. The data was taken from scraping to social media by "bakso malang" keywords. A total of 331 lexicon-filtering Tweets were gathered in the middle of March 2023. The corpus data were analyzed by valance aware dictionary and sentiment reasoner (Vader) sentiment analysis. K-means conducted the cluster of expressions. The result shows the variation of sentiments expressed in social media by anger, disgust, fear, joy, sadness, and surprise with 1.21%, 1.21%, 4.83%, 32.33%, 3.02%, and 57.40%, respectively. Many contents were discussed with keywords of "malang", "bakso", "enak", "makan", "foodfess2", "mau", and "pengen". Behavioural insight shows joy expression when discussing Bakso Malang. Further insights and recommendations are discussed in the paper. [ABSTRACT FROM AUTHOR]

This paper demonstrates a holistic approach for conducting multi-platform and multilingual sentiment analysis of a main stream football club in the United Arab Emirates. The paper reports first iterations of the research project which anticipates to build an automated sentiment analysis system for football clubs to detect and prevent a potential social media crisis at an early stage. The article illustrates a schematic of how state of the art machine learning algorithms, methods and techniques can be put together to provide a basis for automated data collection, classification, sentiment analysis and visual reporting. The article also presents that the time-consuming and costly process of human text labelling can be effectively replaced by lexicon-based sentiment analysis. This paper reports the complete process from data collection to visual reports to discuss challenges and opportunities associated with design and development of such artefacts. [ABSTRACT FROM AUTHOR]

Sentiment analysis is a cognitive tool to extract the emotional tone of a piece of text. It is an area of research that is actively pursued in Natural Language Processing. Social media, forums and blogs, among other places, have seen a massive surge in the number of personalized reviews since the emergence of Internet-based applications. Sentiment analysis is primarily concerned with the classification and prediction of users' thoughts and emotions from these reviews. In recent years, numerous deep learning techniques have emerged to achieve this task. This paper provides a technical summary of Sentiment Analysis using a Bidirectional LSTM network. This model is capable of handling long-term dependencies by introducing memory into the model for making better predictions. The Amazon Product Review dataset served as the source of data for this research. Concretely, the analysis is performed on 104,975 product reviews reflecting users' attitudes toward mobile electronics products. The proposed model attempts to classify the reviews into two categories: positive and negative. Finally, the paper outlines the results of the analysis and suggests potential avenues for future research. [ABSTRACT FROM AUTHOR]

Text sentiment tendency analysis is a hot task in natural language processing. And text as the essential expression form of language, both individual word information, and overall utterance, deserves to be focused on. This paper proposes a fusion model to achieve high precision text sentiment analysis. This model combines the advantages of CNN to extract local information of text and BiLSTM to extract contextual association of text and introduces the attention mechanism to increase the focus on words with a solid emotional tendency in the text. The training datasets are comments that crawled from several social media sites such as Facebook, Twitter, Instagram, WhatsApp, etc. Based on the attention mechanism, this paper investigates the semantic sentiment analysis to reach the study of classification prediction for analyzing the positive and negative sentiment of financial news, social media, etc. The experimental results show that the proposed method can better extract features from the text and classify them than other baseline models. [ABSTRACT FROM AUTHOR]

Generating fake news has become a prevalent issue in the digital age, with profound consequences on society, politics, and media. Due to an exponential increase in the use of unverified information, spread on social media by users with different backgrounds, a significant task facing natural language processing is real-time recognition of generated fake news. The objective of this paper is to investigate the feasibility of utilizing artificial intelligence (AI) for the automatic generation of fake news that possesses a sufficiently high level of quality to serve as a synthetic corpus. These automatically generated texts are subsequently used to improve fake news detectors, evaluated on different publicly available datasets fake news datasets, but also against a novel fake news corpus, especially developed for this research. [ABSTRACT FROM AUTHOR]

The issue of natural disasters, their management, and their prevention, requires good communication among stakeholders and between stakeholders and the population. The Nyiragongo volcanic crisis was a traumatic experience for those who were in Goma city in 2021, on 22 and 23 May. It was characterized by insufficient communication, demonstrating the lack of cohesion between stakeholders. In order to prepare a possible implementation of dedicated social media and groupware for effective communication during the Nyiragongo Volcanic Crisis, this paper presents a survey conducted among Goma citizens and proposes a possible architecture of a dedicated social media integrating text mining, which would help develop a flexible evacuation plan. Based on the results obtained, it appears that the establishment of a single information system related to the volcanic crisis can reduce the challenges related to social media where the volcano observatory, political authorities, different partners and citizens can share alert messages, locations of danger, and more comments. This unified information system would help to gather, discuss and disseminate accurate information on volcanic activities and associated hazards and risks. [ABSTRACT FROM AUTHOR]

Suicidal ideation detection from social media is an evolving research with great challenges. Many of the people who have the tendency to suicide share their thoughts and opinions through social media platforms. As part of many researches it is observed that the publicly available posts from social media contain valuable criteria to effectively detect individuals with suicidal thoughts. The most difficult part to prevent suicide is to detect and understand the complex risk factors and warning signs that may lead to suicide. This can be achieved by identifying the sudden changes in a user's behavior automatically. Natural language processing techniques can be used to collect behavioral and textual features from social media interactions and these features can be passed to a specially designed framework to detect anomalies in human interactions that are indicators of suicidal intentions. We can achieve fast detection of suicidal ideation using deep learning and/or machine learning based classification approaches. For such a purpose, we can employ the combination of LSTM and CNN models to detect such emotions from posts of the users. In order to improve the accuracy, some approaches like using more data for training, using attention model to improve the efficiency of existing models etc. could be done. This paper proposes a LSTM-Attention-CNN combined model to analyze social media submissions to detect any underlying suicidal intentions. During evaluations, the proposed model demonstrated an accuracy of 90.3% and an F1-score of 92.6%, which is greater than the baseline models. [ABSTRACT FROM AUTHOR]

Sarcasm is a complex linguistic phenomenon involving humor, criticism, or phrases that convey the opposite meaning, mask true feelings, and play pivotal roles in various aspects of communication. Therefore, identifying sarcasm is essential for sentiment analysis, social media monitoring, and customer service, as it enables a better understanding of public sentiment. Moreover, social media has become a primary platform for people to express their feelings and opinions and provide feedback to businesses and service providers. Misinterpreting sarcasm in customer feedback can lead to incorrect responses and actions. However, accurately detecting sarcasm is challenging because it depends on context, cultural factors, and inherent ambiguity. Despite the plenty of research and resources in Machine Learning (ML) for detecting sarcasm in English, including Deep Learning (DL) techniques, there is still a shortage of research in sarcasm detection in Arabic, particularly in DL methodologies and available sarcastic datasets. This paper constructed a new Arabic sarcastic corpus and fine-tuned three pre-trained Arabic transformer-based Language Models (LM) for Arabic sarcasm detection. We also proposed a hybrid DL approach for sarcasm detection that combines static and contextualized representations using pre-trained LM, such as Word2Vec word embeddings and Bidirectional Encoder Representations from Transformers (BERT) models pretrained on Arabic resources. The proposed enhanced hybrid deep learning approach outperforms state-of-the-art models by 8% on a shared benchmark dataset and achieves a 5% improvement in F1-score on another. [ABSTRACT FROM AUTHOR]

Abstract Sentiment analysis on social media such as Twitter has become a very important and challenging task. Due to the characteristics of such data—tweet length, spelling errors, abbreviations, and special characters—the sentiment analysis task in such an environment requires a non-traditional approach. Moreover, social media sentiment analysis is a fundamental problem with many interesting applications. Most current social media sentiment classification methods judge the sentiment polarity primarily according to textual content and neglect other information on these platforms. In this paper, we propose a neural network model that also incorporates user behavioral information within a given document (tweet). The neural network used in this paper is a Convolutional Neural Network (CNN). The system is evaluated on two datasets provided by the SemEval-2016 Workshop. The proposed model outperforms current baseline models (including Naive Bayes and Support Vector Machines), which shows that going beyond the content of a document (tweet) is beneficial in sentiment classification, because it provides the classifier with a deep understanding of the task. [ABSTRACT FROM AUTHOR]

The study of Online Social Networks offers growing opportunities to examine a number of aspects of the real world and to better understand how human society works at scale. One crucial research direction, constantly evolving over time, concerns the language adopted by users and its impact on online communities. To provide a contribution in this setting, in this paper, we adopt a multi-relational model for social network and identify a new typology of relation, namely co-interest, as an explicit common declaration of engagement towards a given topic among pair of users. Thematic communities can be, hence, derived leveraging such relations. After that, exploiting Natural Language Processing and Machine Learning techniques, we identify and define suitable analysis metrics focusing on the characteristics of the language adopted in such communities. With the objective of analyzing how the adoption of different language may impact the formation of cohesive and strong communities, to make our study general, we consider two very popular and intrinsically different social platforms, namely Twitter and Reddit. The analysis carried out in this paper aims at comparing the two social media to find interesting and novel results about how their users compose textual comments and the corresponding impact on users' interaction. Interestingly, the obtained results show that, although the main factor is related to the popularity of the user generating the content, the characteristics of the adopted language play a non negligible role in the formation of strong communities in social systems. • Investigation about language characteristics of Online Social Networks posts (OSN). • Language analysis to identify the evolution dynamics of online communities. • Correlation between potential post engagement and user adopted language style. • Relationship between the content produced in OSNs and the formation of communities. [ABSTRACT FROM AUTHOR]

The tourism industry has particularly evolved in the era of social media, and due to the huge amount of data available on the web, tourism forums have become spaces for exchanging information about many topics concerning a country's tourism and hospitality, which has helped to establish a bank of knowledge that might encourage artificial intelligence researchers to follow some research tracks that could lead to interesting insights for decision makers in the sphere of tourism. This paper is an application of artificial intelligence within the tourism industry in the context of Morocco, whereby we carried out natural language processing transformers to classify entities (i.e., Named Entity Recognition) in text that was collected from the Moroccan forum in TripAdvisor. [ABSTRACT FROM AUTHOR]

In recent years, Aspect-based Sentiment Analysis (ABSA) has been a crucial yet challenging task in recognizing textual emotions from text. ABSA has numerous application across various fields, such as social media, commodity review, and movie comment, making it an attractive area of research. Many researchers are working to develop more powerful sentiment analysis models. Currently, most existing ABSA models use the generic pre-trained language models (PLMs) based fine-tuning paradigm, which only utilizes the encoder parameters while discarding the decoder parameters of PLMs. However, this approach fails to leverage the prior knowledge revealed in PLMs effectively. To address these issues, we investigate the potential of the initial pre-training scheme of PLMs to conduct ABSA and thus propose a novel approach in this paper, namely Target Word Transferred ABSA (WordTransABSA). In WordTransABSA, we propose "Word Transferred LM", a novel sequence-level optimization strategy that transferred target words in sentence into pivot tokens to stimulate better PLM semantic understanding capability. Given a sentence with aspect terms as input, WordTransABSA generates contextually appropriate semantics and predicts the affective tokens on the corresponding positions of the aspect terms. The final sentiment polarity of each aspect term is determined through several sentiment identification strategies that we selected. WordTransABSA takes full advantage of the versatile linguistic knowledge of Pre-trained Language Model, resulting in competitive accuracy compared with recent baselines. The WordTransABSA demonstrates its superiority and effectiveness through extensive experiments in both data-sufficient (full-data supervised learning) and data-insufficient (few-shot learning) scenarios. We have made our code publicly available on GitHub: https://github.com/albert-jin/WordTransABSA. • We present WordTransABSA, introducing a "Word Transferred LM" strategy for ABSA. • Various strategies are explored for searching sentiment-related pivot tokens. • Data-sufficient/scarce experiments validate the effectiveness of WordTransABSA. • Stimulating PLM pre-training paradigm is proved to be a better solution for ABSA. [ABSTRACT FROM AUTHOR]

The proliferation of the Internet and social media usage creates enormous textual data (specifically, news content) on the web. The most proportion of contents primarily are unstructured. Extracting meaningful insights from unstructured content is nearly impossible or extremely hard, and time-consuming by human labor. Thus, automatic text classification has gained much attention from NLP experts in recent years. Several techniques have been developed to classify news text in high resource languages (e.g., English, Chinese, French). However, the automatic classification of Bengali news text is in a primitive stage to date. This paper investigates the six most popular machine learning techniques (such as Logistic Regression (LR), Support Vector Classifier (SVC), Decision Tree (DT), Multinomial Naive Bayes (MNB), Random Forest (RF), etc.) with Term Frequency-Inverse Document Frequency (TF-IDF) features for automatic sports news classification in Bengali. Due to the unavailability of benchmark corpus, this work also developed a Bengali news corpus (called BNeC) consisting of 43306 news documents with 202830 unique words in multiple classes: Cricket, Football, Tennis, and Athletics. Experimental results on the test dataset show that the Support Vector Classifier (SVC) with unigram+bigram+trigram feature space obtained the highest weighted f1-score of 97.60% than the other classifiers and feature combinations. [ABSTRACT FROM AUTHOR]

Classifying textual emotion plays a critical role in several HCI applications where the text is utilized as a central means of communication such as messages, reviews, blogs and other Web 2.0 platforms. The extensive usage of the Internet has emerged as an unprecedented means for people to express their feelings or emotion on blogs, social media, and e-commerce sites in recent years. Most of the emotions displayed on the online platforms are in textual forms (such as posts, tweets, comments and reviews), which are unorganized and time-consuming to structured due to their disordered forms. Although several emotion analysis tools are available in high-resource languages, it is critical to developing an automatic emotion classification system for low-resource languages, including Bengali, due to its constrained resources. This paper presents an ML-based ensemble method to classify six primary textual emotions (anger, fear, disgust, sadness, surprise and joy) from Bengali texts. An emotion corpus containing 8047 Bengali texts is developed to perform the textual emotion classification task. This work investigates eight standard ML-based techniques such as logistic regression (LR), multinomial naive Bayes (MNB), support vector machine (SVM), random forest (RF), decision tree (DT), K-nearest neighbour (KNN) and adaptive boosting (AdaBoost) and an ensemble method (a combination of LR, RF, SVM) with Bag of words (BoW) and tf-idf feature extraction techniques. The experimental result demonstrates that the ensemble with tf-idf achieved the highest weighted f 1 -score of 62.39% compared to other methods. [ABSTRACT FROM AUTHOR]

Natural Language Processing (NLP) applications such as text categorization, machine translation, sentiment analysis, etc., need annotated corpora and lexicons to check quality and performance. This paper describes the development of resources for sentiment analysis specifically for Arabic text in social media. A distinctive feature of the corpora and lexicons developed are that they are determined from informal Arabic that does not conform to grammatical or spelling standards. We refer to Arabic social media content of this sort as Dialectal Arabic (DA) - informal Arabic originating from and potentially mixing a range of different individual dialects. The paper describes the process adopted for developing corpora and sentiment lexicons for sentiment analysis within different social media and their resulting characteristics. The addition to providing useful NLP data sets for Dialectal Arabic the work also contributes to understanding the approach to developing corpora and lexicons. [ABSTRACT FROM AUTHOR]

The usage of digital/social media is increasing day by day with the advancement of technology. People in the twenty-first century are being raised in an internet-enabled world with social media. Communication has been just one button click. Even though there are plenty of opportunities with digital media people tend to misuse it. People spread hatred toward a person in social networking. Cyberbullying affects people in different aspects. It doesn't affect only for health, there are more different aspects which will lead life to a threat. Cyberbullying is a worldwide modern phenomenon which humans cannot avoid hundred percent but can be prevented. Most existing solutions have shown techniques/approaches to detect cyberbullying, but they are not freely available for end-users to use. They haven't considered the evolution of language which makes a big impact on cyberbullying text. This paper describes a system for automatic detection and prevention cyberbullying considering the main characteristics of cyberbullying such as Intention to harm an individual, Repeatedly and over time and using abusive curl language or hate speech using supervised machine learning. The system relies on the detection of cyberbullying text along with the themes/categories associated with cyberbullying such as racist, sexual, physical mean, swear and other, using support vector machines and Logistic regression. The author of this research presents a new hypothesis for cyberbullying detection that the circumstances and usage of texting and its language have changed by time. Most of the studies have considered calling someone stupid, ugly and idiot as cyberbullying. Things have changed, such words may or may not always be a bullying incident. If a person wants intentionally to harm an individual, they will use extreme words. In addition to traditional feature extraction techniques like Term Frequency–Inverse Document Frequency (TF-IDF), N-gram and profanity along with sentiment analysis increases the accuracy of the system. Evaluated the proposed system using Recall, Precision and F1-score. [ABSTRACT FROM AUTHOR]

During the COVID-19 pandemic, online social media platforms such as Twitter facilitate the exchange of information among people. However, the prevalence of "infodemic" such as online hate speech has exacerbated social rifts, discrimination, prejudice and even hate crimes. Timely and effective detection of the hate speech will help create a healthy public opinion environment. Most of the current COVID-19-related hate speech research focuses on a single language, such as English. In this paper, we introduce a cross-lingual transfer learning method, aiming to contribute to hate speech detection in low-resource languages. We propose a deep learning based model to classify hate speech with a pre-trained language model for multilingual text embedding. Data augmentation and cross-lingual contrastive learning are then utilized to further improve the performance of cross-lingual knowledge transfer. To evaluate our method, we collected three publicly available annotated COVID-19-related hate speech datasets on Twitter, i.e., two in English and one in German. Furthermore, a Chinese dataset based on Weibo is constructed to expand multilingual data. The experimental results across three languages illustrate the effectiveness of our method for cross-lingual hate speech detection. Test F1-scores of our method for English, Chinese, German as transfer target languages can reach up to 0.728, 0.799 and 0.612 respectively, which are on average better than other baselines. • Suggested a cross-lingual COVID-19-related hate speech detection problem. • Introduced cross-lingual contrastive learning. • Validated model performance on datasets in three languages. [ABSTRACT FROM AUTHOR]

Regret is a universal emotion that arises from sadness, disappointment, or remorse about something that has occurred or that one has done or failed to do in the past. It typically involves wishing that a different decision had been made or that a different action had been taken. Although regret has been studied in various contexts such as psychology, neuroscience, and philosophy, its expression and analysis on social media using natural language processing techniques is a relatively recent research topic that warrants further investigation. In this paper, we present a study of regret and its expression on social media platforms. Specifically, we present a novel dataset of Reddit texts that have been classified into three classes: Regret by Action, Regret by Inaction, and No Regret. We then use this dataset to investigate the language used to express regret on Reddit and to identify the domains of text that are most commonly associated with regret. Our findings show that Reddit users are most likely to express regret for past actions, particularly in the domain of relationships. We also found that deep learning models using GloVe embedding outperformed other models in all experiments, indicating the effectiveness of GloVe for representing the meaning and context of words in the domain of regret. Overall, our study provides valuable insights into the nature and prevalence of regret on social media, as well as the potential of deep learning and word embeddings for analyzing and understanding emotional language in online text. These findings have implications for the development of natural language processing algorithms and the design of social media platforms that support emotional expression and communication. • Regret is an emotion that has a significant impact on behavior and decision-making. • Regret is often triggered by actions or inactions in the past. • Reddit's users often regret their actions about relationships in the past. • Effectiveness of GloVe in representing the meaning and context of words. [ABSTRACT FROM AUTHOR]

Abstract Recently, the vast use of social media and the high availability of internet access have produced a considerably different textual data from the formal and standard data on the Web. This includes various Arabic dialectal languages, which are the native spoken languages of Arabic speakers. The presence of textual Arabic dialectal languages on the Web has brought many new opportunities as well as challenges for machine learning and Arabic language processing. The identification of this type of informal data has its crucial effect on several applications such as sentiment analysis and machine translation. However, the standard NLP tools developed for traditional data fall short due to nature of dialectal textual data. Deep learning tools have proven to be very effective in processing social Media dialectal text. In this paper, we consider a variety of deep learning models for the automatic classification of Arabic dialectal text. We use a free large manually-annotated dataset known as Arabic Online Commentary (AOC), which includes several Dialectal Arabic (DA) along with the Modern Standard Arabic (MSA), [3]. We consider the most frequent dialects in the dataset. Namely, the Egyptian (EGP), Levantine (LEV), and Gulf –including Iraqi - (GLF). Four different deep neural network models have been implemented to examine the Arabic dialectal classification problem for each pair of the 3 dialects (binary classification experiments) as well as one ternary-classification experiment including all dialects together. The results show a varying but promising performance of the models for each pair of dialects. Furthermore, a closer examination on the manually-annotated AOC dataset has been carried out and hence, we conclude that there is a serious demand for a thorough refinement and review of the AOC annotated sentences as it is an important benchmark dataset in the field. [ABSTRACT FROM AUTHOR]

A far-reaching expansion of advanced information technology enables ease and seamless communications over online social networks, which have been a de facto premium correspondents in the current cyber world. The ever-growing social network data has gained attention in recent years and can be handy for industrial revolution 4.0. With the integration of social networks with the Internet of Things being noticed in different industries to enhance human involvement and increase their productivity, security in such networks is increasingly alarming. Vulnerabilities can be characterized in the form of privacy invasion, leading to hazardous contents, which can be detrimental to social media actors and in turn impact the processes of the overall Social Network-Integrated Industrial Internet of Things (SN-IIoT) ecosystem. Despite this prevalence, the current platforms do not have any significant level of functionality to capture, process, and reveal unhealthy content among the social media actors. To address those challenges by detecting hazardous contents and create a stable social internet environment within IIoT, a statistical learning-enabled trustworthy analytic tool for human behaviors has been developed in this paper. More specifically, this paper proposes a machine learning (ML)-enabled scheme SPY-BOT, which incorporates a hybrid data extraction algorithm to perform post-filtering that arbitrates the users' behavior polarity. The scheme creates class labels based on the featured keywords from the decision user and classifies suspicious contacts through the aid of ML. The results suggest the potential of the proposed approach to classify the users' behavior in SN-IIoT. [ABSTRACT FROM AUTHOR]

• Three datasets are merged to build a larger, more diversified spam dataset. • Data balancing is done using SmoteTomek and NearMiss data resampling techniques. • Proposed Hard and Soft voting ensemble models for performance improvement. • Several DL approaches are experimented with GloVe and FastText word embeddings. • Proposed FCB-SA hybrid DL approach outperformed other DL approaches. Cybercriminals use social media platforms to disseminate spam, misleading facts, fake news, and malicious links. Blocking such deceptive social media spam is essential. However, extracting relevant features from social networks is challenging due to privacy and time constraints. Traditional frequency-based word representation techniques are time-consuming and inefficient in producing contextual word vectors. Word embeddings and deep learning models have recently shown good results in text classification. Also, most existing approaches assumed balanced class distribution, which is false for most real-world datasets. In this paper, an attempt is made to advance the performance of the social spam detection system by leveraging dataset balancing, advanced word embedding techniques, machine learning, and deep learning approaches with the self-attention mechanism. In the proposed framework, the datasets are balanced using NearMiss and SmoteTomek techniques to feed several machine-learning models. Later, the baseline ML models and proposed voting-based ensemble models are evaluated on imbalanced and balanced datasets. For the proposed deep learning-based hybrid approaches, embeddings are generated using GloVe and FastText word embeddings on the balanced combined dataset and passed into the deep neural network comprised of Conv1D and Bi-directional recurrent neural network layers with the self-attention mechanism for improved context understanding and effective results. This study examines hybrid approaches for detecting social spam using imbalanced social network data and picks the optimum combination. Besides, Machine learning ensembles, word embeddings, deep learning with hyper-parameter optimization, and a self-attention method are compared thoroughly. Experiments and comparisons with other techniques show that the proposed hybrid framework with deep learning-based approaches achieves better performance. [ABSTRACT FROM AUTHOR]

Over the years, there has been a rise in the number of fabricated and fake news stories that utilize both textual and visual information formats. This coincides with the increased likelihood that users will acquire their news from websites and social media platforms. While there has been various research into the detection of fake news in text using machine learning techniques, less attention has been paid to the problem of multimedia data fabrication. In this paper, we propose a Stylometric, and Semantic similarity oriented for Multimodal Fake News Detection (SSM). There are five distinct modules that make up our methodology: Firstly, we used a Hyperbolic Hierarchical Attention Network (Hype-HAN) for extracting stylometric textual features. Secondly, we generated the news content summary and computed the similarity between Headline and summary. Thirdly, semantic similarity is computed between visual and textual features. Fourthly, images are analyzed for forgery. Lastly, the extracted features are fused for final classification. We have tested SSM framework on three standard fake news datasets. The results indicated that our suggested model has outperformed the base line and state-of-the-art methods and is more likely to detect fake news in complex environments. [ABSTRACT FROM AUTHOR]

Abusive language is an expression (both oral or text) that contains abusive/dirty words or phrases both in the context of jokes, a vulgar sex conservation or to cursing someone. Nowadays many people on the internet (netizens) write and post an abusive language in the social media such as Facebook, Line, Twitter, etc. Detecting an abusive language in social media is a difficult problem to resolve because this problem can not be resolved just use word matching. This paper discusses a preliminaries study for abusive language detection in Indonesian social media and the challenge in developing a system for Indonesian abusive language detection, especially in social media. We also built reported an experiment for abusive language detection on Indonesian tweet using machine learning approach with a simple word n-gram and char n-gram features. We use Naive Bayes, Support Vector Machine, and Random Forest Decision Tree classifier to identify the tweet whether the tweet is a not abusive language, abusive but not offensive, or offensive language. The experiment results show that the Naive Bayes classifier with the combination of word unigram + bigrams features gives the best result i.e. 70.06% of F1 - Score. However, if we classifying the tweet into two labels only (not abusive language and abusive language), all classifier that we used gives a higher result (more than 83% of F1 - Score for every classifier). The dataset in this experiment is available for other researchers that interest to improved this study. [ABSTRACT FROM AUTHOR]

Most of the queries submitted to search engines are composed of keywords but it is not enough for users to express their needs. Through verbose natural language queries, users can express complex or highly specific information needs. However, it is difficult for search engine to deal with this type of queries. Moreover, the emergence of social medias allows users to get opinions, suggestions or recommendations from other users about complex information needs. In order to increase the understandability of user needs, tasks as the CLEF Social Book Search Suggestion Track have been proposed from 2011 to 2016. The aim is to investigate techniques to support users in searching for books in catalogs of professional metadata and complementary social media. In this context, we introduce in the current paper a statical approach to deal with long verbose queries in Social Information Retrieval (SIR) by taking Social Book Search(SBS) as a study case. firstly, a morphosyntactic analysis was introduced to reduce verbose queries, the second step is based on expanding the reduced queries using association rules mining combined with Pseudo relevance feedback. Experiments on SBS 2014 and 2016 collections show significant improvement in the retrieval performance. [ABSTRACT FROM AUTHOR]

Nowadays, fake news has turned into a major problem because of the negative impact that it creates on society. Social media allows people to spread information on the internet with slight investigations and to add fewer filters than the actual content. Nowadays, the false news in the internet community is unsure, and it creates a wrong impression among the users. Detecting false news has become a critical task based on shared content. To tackle the false news growth in social media, various automatic detection schemes were evaluated. The "Natural Language Processing (NLP)" method also gives a prominent solution for false news detection. The main intention of this paper is to design and introduce an innovative false news recognition method using Meta-heuristic Searched-Ensemble Learning (MS-EL). Further, the selected features are extracted by the "Term Frequency-Inverse Document Frequency (TF-IDF)" and also Word2vec features. Here, the extracted selected features are integrated with the Hybrid Squirrel–Dragonfly Search Optimization (HS-DSO) is used to optimize the weighted feature selection approach with the fitness function of solving data variance and correlation. The proposed MS-EL is adopted in the classification part, having three sets of classifiers, Long Short-Term Memory (LSTM), Support Vector Machine (SVM), and Deep Neural Network (DNN). Here, the ensemble classifier is enhanced by the same HS-DSO that shows the parameter tuning with a high convergence rate. From the experimental outcomes, the accuracy of HS-DSO-MS-EL is 22% higher than BMO-MS-EL, 24% higher than SP-BMO-MS-EL, 30% higher than SSA-MS-EL, and 29% higher than DA-MS-EL. Thus, the experimental analysis with standard datasets establishes that the introduced fake news detection method has gained higher accuracy than the existing models. [ABSTRACT FROM AUTHOR]

Internet public social media and forums provide a convenient channel for people concerned about public health issues, such as COVID-19, to share and discuss information/misinformation with each other. In this paper, we propose a natural language processing (NLP) method based on Bidirectional Long Short-Term Memory (Bi-LSTM) technique to perform sentiment classification and uncover various issues related to COVID-19 public opinions. Bi-LSTM is an improved version of conventional LSTMs for generating the output from both left and right contexts at each time step. We experimented with real datasets extracted from Twitter and Reddit social media platforms, and our experimental results showed improved metrics compared with the conventional LSTM model as well as recent studies available in the literature. The proposed model can be used by official institutions to mitigate the effects of negative messages and to understand peoples' concerns during the pandemic. Furthermore, our findings shed light on the importance of using NLP techniques to analyze public opinion and to combat the spreading of misinformation and to guide health decision-making. • Bi-LSTM sentiment detection model is developed to classify COVID-19 social comments. • Various scenarios to help counter negative COVID-19 comments and reduce their impact. • Comparative analysis of our approach with most popular models is presented. • Word embedding followed by Bi-LSTM provides better performance on Sentiment Analysis. [ABSTRACT FROM AUTHOR]

As the prevalence of social media on the Internet, opinion mining has become an essential approach to analyzing so many data. Various applications appear in a wide range of industrial domains. Meanwhile, opinions have diverse expressions which bring along research challenges. Both of the practical demands and research challenges make opinion mining an active research area in recent years. In this paper, we present a review of Natural Language Processing (NLP) techniques for opinion mining. First, we introduce general NLP techniques which are required for text preprocessing. Second, we investigate the approaches of opinion mining for different levels and situations. Then we introduce comparative opinion mining and deep learning approaches for opinion mining. Opinion summarization and advanced topics are introduced later. Finally, we discuss some challenges and open problems related to opinion mining. [ABSTRACT FROM AUTHOR]

This paper presents a supervised learning method for irony detection in Arabic tweets. A binary classifier uses four groups of features whose efficiency has been empirically proved in other languages such as French, English, Italian, Dutch and Japanese. Our first results are encouraging and show that state of the art features can be successfully applied to Arabic language with an accuracy of 72.76%. [ABSTRACT FROM AUTHOR]

There has been an enormous increase in the volume of Arabic posts on many social networks, providing a rich source of opinions on a variety of topics. Opinion mining has attracted much research interest in the last decade, but has mostly been conducted using the English language. Opinion mining from Arabic social media is lagging behind, primarily due to challenges with processing the morphologically complex Arabic natural language and the lack of available tools and resources for extracting Arabic sentiments from text. This complex task is further exacerbated when dealing with Saudi dialects that do not abide by the formal grammatical structure of Modern Standard Arabic (MSA). Targeting the use-case of unemployment in Saudi Arabia, this paper investigates the major challenges facing sentiment analysis researchers of the informal Arabic language and documents how these challenges can be initially addressed by means of linguistic pre-processing of raw text and supervised machine learning (ML) to identify sentiment polarity. The initial experimental evaluation yielded satisfactory sentiment classification results while also demonstrating the benefit of using ML to identify the dominant features in deciding sentiment polarity, which can in turn be used in constructing a problem-domain knowledge base that can further improve the accuracy of sentiment classification. [ABSTRACT FROM AUTHOR]

Twitter, is considered as one of the famous social networking platform. It has become a very valuable information source for many Natural Language Processing (NLP) applications. Some strategies and linguistic pipelines were developed for analyzing English tweets but Arabic social media analysis is still an active research area. In this research paper, we focus on the task of pre-processing Arabic tweets, which can be regarded as a first step for any NLP application. We follow up with a statistical machine translation for Arabic tweets into English, where we explain the normalization process for both Arabic and English tweets. Moreover, to overcome the obstacle of unavailability of Arabic-English parallel corpora in the social media context, we used the UN corpus, a more general corpus in (Modern Standard Arabic and English). Then, we applied adapting strategies for the tweet’s contents like using an out-of-domain and/or in-domain language model. Our conducted experiments showed that applying a good lexical normalization on both languages and combining in-domain and out-of-domain data for the language model improves the Bleu score with 4pt., over the baseline. [ABSTRACT FROM AUTHOR]

Sentiment analysis (SA) is an intellectual process of extricating user's feelings and emotions. It is one of the pursued field of Natural Language Processing (NLP). The evolution of Internet based applications has steered massive amount of personalized reviews for various related information on the Web. These reviews exist in different forms like social Medias, blogs, Wiki or forum websites. Both travelers and customers find the information in these reviews to be beneficial for their understanding and planning processes. The boom of search engines like Yahoo and Google has flooded users with copious amount of relevant reviews about specific destinations, which is still beyond human comprehension. Sentiment Analysis poses as a powerful tool for users to extract the needful information, as well as to aggregate the collective sentiments of the reviews. Several methods have come to the limelight in recent years for accomplishing this task. In this paper we compare the various techniques used for Sentiment Analysis by analyzing various methodologies. [ABSTRACT FROM AUTHOR]

Large amount of data is being generated from sensors, satellites, social media etc. This big data (velocity, variety, veracity, value and veracity) can be processed so as to make timely decisions by the decision makers. This paper presents results of the proposed Hadoop framework that performs entity resolution in Map and reduce phase. MapReduce phase matches two real world objects and generates rules. The similarity score of these rules are used for matching stream data during testing phase. Similarity is calculated using 13 different semantic measures such as token-based similarity, edit-based similarity, hybrid similarity, phonetic similarity as well as domain dependent Natural language processing measures. Semantic measures are implemented using Hive programming. The proposed system is tested using e-catalogues of Amazon and Google. [ABSTRACT FROM AUTHOR]

Sentiment analysis or opinion mining is the process of computationally identifying and categorizing opinions expressed in a piece of text, in order to determine whether the writer's attitude towards a particular topic, product, etc. is positive, negative, or neutral. It is one of the most active research areas in natural language processing and text mining in recent years. A detailed study of the two concepts (1) Temporal sentiment analysis (2) Sentiment causal relation is presented in this paper. Temporal sentiment analysis is useful for summarizing the events based on sentiment and time. Causal relation is useful for identifying cause and effect of events and is also useful for event prediction. These two concepts when combined result in a better event prediction model that can predict the time period between the events and sentiment of upcoming events. The proposed work introduces a generalized prediction model based on temporal sentiment analysis of tweet to identify the causal relation between the events which can be used to predict the event sentiment and duration between the events. The proposed method is to be evaluated using the performance measures precision and recall. The accuracy of causal rule prediction is evaluated using parameters Mean Absolute Error (MAE) and the Root Mean Squared error (RMSE). [ABSTRACT FROM AUTHOR]

Sentiment Analysis is an area of study within Natural Language Processing that is concerned with identifying the mood or opinion of subjective elements within a text. This paper focuses on the various methods used for classifying a given piece of natural language text according to the opinions expressed in it i.e. whether the general attitude is negative or positive. We also discuss the two-step method (aspect classification followed by polarity classification) that we followed along with the experimental setup. [ABSTRACT FROM AUTHOR]

These days social media services are widespread and are infrastructure of communication in the Internet. Since Twitter is one of the most famous real-time communication services, we can grasp opinions of crowds in the real world analyzing tweets. There are usually various kind of opinions in Twitter and you need to deal with the opinions carefully. In this paper we focus on tweets on an incident and extract tweets reflecting sufferers’ opinions. When a incident happens, vast amount of tweets are created by many Twitter users. We compare tweets by sufferers with ones by others and extract tweets unique to the sufferers with density ratio estimation. In experiments we confirm that our proposed method can extract tweets including sufferers’ opinions. [ABSTRACT FROM AUTHOR]

• Geotagged photographs are used to investigate cultural ecosystem services (CESs) to the coastal areas of Hong Kong. • Visiting sentiments are accessed through user-generated textual content as a proxy in the analysis of the geotagged photographs. • Both cultural attractions and protected areas are typical hotspots associated with CESs. • The relationships between CESs-related sentiments and critical sociodemographic and temporal factors are discussed. Ecosystem service assessments have been conducted for coastal management with a primary focus on tangible natural capital. Thus far, the explicit variation of cultural ecosystem services (CESs) and particularly the sentiments of CES-related ecotourism are not well understood. This paper takes advantage of big social media data to unravel the patterns of CESs and visiting sentiments in coastal areas of Hong Kong. Through machine-learned keyword labels, we employed content analysis to derive visual information for geotagged photographs. Applying natural language processing to apps with machine learning in cloud computing, we derived the sentiments based on user-generated textual content associated with coastal ecotourism. Based on regression analysis and multiple comparisons analysis, we identify the association between critical demographic and temporal factors with CES and related visiting sentiments. Referring to previous studies, we identified the main intangible benefits into six basic divisions based on 424 keyword labels for coastal areas. Our results show that hotspots of CESs are spatially concentrated in both cultural attractions and protected areas, which are critical for coastal ecosystem management and protection. More specifically, these areas of high CES value have good spatiotemporal accessibility, high infrastructure coverage, and spatially explicit population and economic growth. Furthermore, we discover that sentiments related to coastal CESs vary based on social media characteristics. Our study renews the indicators of quantitative CES evaluation based on crowdsourcing geospatial data. [ABSTRACT FROM AUTHOR]

Abstract: The research described in this paper is focused on analyzing two playful domains of language: humor and irony, in order to identify key values components for their automatic processing. In particular, we are focused on describing a model for recognizing these phenomena in social media, such as “tweets". Our experiments are centered on five data sets retrieved from Twitter taking advantage of user-generated tags, such as “#humor" and “#irony". The model, which is based on textual features, is assessed on two dimensions: representativeness and relevance. The results, apart from providing some valuable insights into the creative and figurative usages of language, are positive regarding humor, and encouraging regarding irony. [Copyright &y& Elsevier]

Social media platforms have radically transformed the creation and dissemination of news. Users can easily access this news in a fast and efficient manner. However, some users might post negative and fraudulent content in the form of comments or posts. Such content can constitute a threat to an individual or an organization. Therefore, the identification of fake news has become a major research field in natural language processing (NLP). The main challenge is to determine whether the news is real or fake. In this paper, we propose an attention-based convolutional bidirectional long short-term memory (AC-BiLSTM) approach for detecting fake news and classifying them into six categories. The evaluation of our proposed approach on a benchmarked dataset shows a significant improvement in accuracy rate in comparison with other existing classification models. In particular, this work contributes to the progress in the field of detecting fake news and confirms the feasibility of our proposed approach in classifying fake news on social media. • Tackles fake news detection problem in a multi-class environment. • Employed pre-trained word embedding for representing input sequences. • Proposed attention-based convolutional bidirectional recurrent neural networks. • Improved the accuracy of fake news detection. [ABSTRACT FROM AUTHOR]

Financial technology-based firms (FinTech) are crucial to promoting new technologies and advances in innovations related to the financial field, sustainable development, and financial inclusion. This paper aims to assess the Sustainable Development Goals (SDGs) and open innovation integrated within the public discourse in the social media of FinTech firms, using machine-learning-based social media analysis (SMA). Accordingly, we tracked the behavior of 21 firms based on the empirical material of 32,716 posts on Twitter. The outcomes showed dissimilar discourses based on FinTech firms' activities. However, it was found that only financial infrastructure, lending, and personal finance fields have a discourse related to innovation and, to a lesser extent, related to the SDGs thematic. Thus, awareness of sustainable objectives is still far from being a relevant issue for FinTech, which, in general, has neglected to significantly mention SDGs; nonetheless, innovation and related terms are a consistent topic in this area. Furthermore, hints of the implementation of the open innovation paradigm and interest in novel technologies are demonstrated, in addition to the promotion of different actors and events on social media that serve as a showcase for firms that have a presence on Twitter. [ABSTRACT FROM AUTHOR]

Twenty-four studies on twenty-three distinct languages and eleven social media illustrate the steady interest in deep learning approaches for multilingual sentiment analysis of social media. We improve over previous reviews with wider coverage from 2017 to 2020 as well as a study focused on the underlying ideas and commonalities behind the different solutions to achieve multilingual sentiment analysis. Interesting findings of our research are (i) the shift of research interest to cross-lingual and code-switching approaches, (ii) the apparent stagnation of the less complex architectures derived from a backbone featuring an embedding layer, a feature extractor based on a single CNN or LSTM and a classifier, (iii) the lack of approaches tackling multilingual aspect-based sentiment analysis through deep learning, and, surprisingly, (iv) the lack of more complex architectures such as the transformers-based, despite results suggest the more difficult tasks requires more elaborated architectures. • Review of applications of Deep Learning to tackle Multilingual Sentiment Analysis. • Fast-growing interest in this field, 24 related papers since 2017 to 2020. • Coverage of 23 different languages and 11 social media data or corpus. • Mixed performance, but word embeddings and CNN or LSTM as trending choices. • Embeddings>feature extractor>classifier, prevailing architecture except for aspect SA. [ABSTRACT FROM AUTHOR]

Undoubtedly, social media, such as Facebook and Twitter, constitute a major part of our everyday life due to the incredible possibilities they offer to their users. However, Twitter and generally online social networks (OSNs) are increasingly used by automated accounts, widely known as bots, due to their immense popularity across a wide range of user categories. Their main purpose is the dissemination of fake news, the promotion of specific ideas and products, the manipulation of the stock market and even the diffusion of sexually explicit material. Therefore, the early detection of bots in social media is quite essential. In this paper, two methods are introduced targeting this that are mainly based on Natural Language Processing (NLP) to distinguish legitimate users from bots. In the first method, a feature extraction approach is proposed for identifying accounts posting automated messages. After applying feature selection techniques and dealing with imbalanced datasets, the subset of features selected is fed in machine learning algorithms. In the second method, a deep learning architecture is proposed to identify whether tweets have been posted by real users or generated by bots. To the best of the authors' knowledge, there is no prior work on using an attention mechanism for identifying bots. The introduced approaches have been evaluated over a series of experiments using two large real Twitter datasets and demonstrate valuable advantages over other existing techniques targeting the identification of malicious users in social media. [ABSTRACT FROM AUTHOR]

Adverse drug reaction events have become one of the main causes of patient death. Since traditional post-marketing surveillance systems based on spontaneous reports have a serious underreporting issue, in recent years research on the detection of adverse reaction events using social media such as Twitter as a data source has attracted increasing attention in recent year. Deep learning models usually rely on a large number of training samples. However, due to the characteristics of user-generated content and the time-consuming data annotation process, related research is faced with the problems caused by small-scale annotated datasets, which restricts deep learning models in achieving satisfactory results. Accordingly, we introduce two regularization methods are introduced at the representation level, i.e., graph embedding-based data augmentation and adversarial training, to improve the performance of detecting adverse events under such conditions. Besides, the applicable scope of these two methods is analyzed and discussed through experiments. Combined with the convolutional neural network, this paper proposes an adverse drug event detection framework that can make full use of the methods. [ABSTRACT FROM AUTHOR]

• New method to semi-automatically build labeled stance detection datasets from Twitter. • Translation strategies outperform zero-shot approaches when data is translated to a high-resourced language. • User-based information helps to label individual tweets. • Our method is applicable to quickly and cheaply generate labeled Twitter-based data. Popular social media networks provide the perfect environment to study the opinions and attitudes expressed by users. While interactions in social media such as Twitter occur in many natural languages, research on stance detection (the position or attitude expressed with respect to a specific topic) within the Natural Language Processing field has largely been done for English. Although some efforts have recently been made to develop annotated data in other languages, there is a telling lack of resources to facilitate multilingual and crosslingual research on stance detection. This is partially due to the fact that manually annotating a corpus of social media texts is a difficult, slow and costly process. Furthermore, as stance is a highly domain- and topic-specific phenomenon, the need for annotated data is specially demanding. As a result, most of the manually labeled resources are hindered by their relatively small size and skewed class distribution. This paper presents a method to obtain multilingual datasets for stance detection in Twitter. Instead of manually annotating on a per tweet basis, we leverage user-based information to semi-automatically label large amounts of tweets. Empirical monolingual and cross-lingual experimentation and qualitative analysis show that our method helps to overcome the aforementioned difficulties to build large, balanced and multilingual labeled corpora. We believe that our method can be easily adapted to easily generate labeled social media data for other Natural Language Processing tasks and domains. [ABSTRACT FROM AUTHOR]

Nowadays, due to the great uncontrolled content posted daily on the Web, there has also been a huge increase in the dissemination of hate speech worldwide. Social media, blogs and community forums are examples where people are freely allowed to communicate. However, freedom of expression is not always respectful since offensive or insulting language is sometimes used. Social media companies often rely on users and content moderators to report on this type of content. Nevertheless, due to the large amount of content generated every day on the Web, automatic systems based on Natural Language Processing techniques are required for identifying abusive language online. To date, most of the systems developed to combat this problem are mainly focused on English content, but this issue is a worldwide concern and therefore other languages such as Spanish are involved. In this paper, we address the task of Spanish hate speech identification on social media and provide a deeper understanding of the capabilities of new techniques based on machine learning. In particular, we compare the performance of Deep Learning methods with recently pre-trained language models based on Transfer Learning as well as with traditional machine learning models. Our main contribution is the achievement of promising results in Spanish by applying multilingual and monolingual pre-trained language models such as BERT, XLM and BETO. • Addressing hate speech detection on social media for Spanish. • Comparing the performance of Deep Learning and Machine Learning models. • Evaluating recent pre-trained language models based on Transformer mechanism. • Comparing pre-trained multilingual and monolingual models available for Spanish. • Conducting an error analysis in order to understand the difficulty of the task. [ABSTRACT FROM AUTHOR]