Twitter Dataset Research Articles

Robust Drug Use Detection on X: Ensemble Method with a Transformer Approach

There is a growing trend for groups associated with drug use to exploit social media platforms to propagate content that poses a risk to the population, especially those susceptible to drug use and addiction. Detecting drug-related social media content has become important for governments, technology companies, and those responsible for enforcing laws against proscribed drugs. Their efforts have led to the development of various techniques for identifying and efficiently removing drug-related content, as well as for blocking network access for those who create it. This study introduces a manually annotated Twitter dataset consisting of 112,057 tweets from 2008 to 2022, compiled for use in detecting associations connected with drug use. Working in groups, expert annotators classified tweets as either related or unrelated to drug use. The dataset was subjected to exploratory data analysis to identify its defining features. Several classification algorithms, including support vector machines, XGBoost, random forest, Naive Bayes, LSTM, and BERT, were used in experiments with this dataset. Among the baseline models, BERT with textual features achieved the highest F1-score, at 0.9044. However, this performance was surpassed when the BERT base model and its textual features were concatenated with a deep neural network model, incorporating numerical and categorical features in the ensemble method, achieving an F1-score of 0.9112. The Twitter dataset used in this study was made publicly available to promote further research and enhance the accuracy of the online classification of English-language drug-related content.

Using social networking evidence to examine the impact of environmental factors on social Followings: An innovative Machine learning method

Social media platforms have changed the way of communication. They are making the world more connected and provide information more quickly, which has resulted in beneficial and harmful effects on society. Analyzing data from multiple social media platforms to spot patterns and trends in user behavior in response to environmental changes is part of using social networking evidence to investigate the effects of environmental variables.The infinite collection of personal data generates privacy issues and ethical questions about data security and utilization.We present a novel framework called meta-heuristic atom search optimized dynamic artificial neural network (MASO-DANN) method for examining the impact of environmental factors using social networking evidence and the complementary abilities between MASO as well as DANN. The study aims to determine the connections between the dynamics of users' social relationships and involvement by examining the data from various online social platforms. We collected a Twitter dataset and preprocessed the data using the stop word removal preprocessing method.With the use of MASO-DANN, extensive social networking data analysis has been rendered achievable, which offers an exhaustive overview of the complex dynamics influencing online communities. The result findings examined various metrics, such as accuracy (95.07%), precision (89%), recall (88%), F1-score (90%) and RMSE (1.6).These metrics demonstrate the robustness along with the dependability of MASO-DANN and its capacity to provide more accurate predictions that evaluate the effects of environmental factors based on social networking evidence.

Geographical Tweets Based Current Affairs Prediction Using Hybrid Features

Twitter serves as a prominent and freely accessible social networking platform, enabling registered and authorized users to share their opinions and reviews through concise messages known as tweets. This research aims to conduct sentiment analysis on a Twitter dataset, seeking to predict and analyze current affairs based on user behaviour and opinions. The scope of current affairs encompasses a wide range of topics, including product reviews, political discourse, movie ratings, and more, expressed in both real-time streaming and offline data on the Twitter platform. During the feature extraction phase, a hybrid approach is employed, incorporating features such as acronyms, synonyms, and emoticons utilized by users in their tweets. Additionally, a hybrid dictionary is introduced, and instead of relying solely on unigrams and bigrams, a novel algorithm, n-gram, is implemented for more comprehensive analysis. This paper delves into various methodologies, specifically focusing on multiclass classification, and introduces a proposed system along with its noteworthy results. The research contributes to the understanding of user sentiments and behaviours on Twitter, offering valuable insights into diverse current affairs.

Classifying Opinions and Sentiments on Social Networking Sites using Machine Learning Classifiers

Abstract: People now publish evaluations on social media for any product, movie, or location they visit as a result of the Web's rapid development. Customers and product owners can both benefit from the reviews posted on social media in order to assess their offerings. Compared to unstructured data, structured data is simpler to analyze. The reviews are mostly available in an unstructured format. Aspect-Based Sentiment Analysis extracts from the reviews the features of a product and then calculates sentiment for each feature. Sentiment analysis, also known as opinion mining, is a natural language processing (NLP) technique used to determine the sentiment or emotional tone expressed in a piece of text, such as a sentence, paragraph, or document. Machine leaning classifiers are used to classify sentiments. Machine learning classifiers cannot process raw text so raw text needs to be converted into vectorized form. Feature extraction techniques are used to convert raw text to numerical form also called vectorized data. In present research, four feature extraction techniques with five different machine learning classifiers namely, SVM, Logistic Regression, Naïve Bayes, Random Forest, and KNN are applied to classify sentiments associated with tweets. Two online twitter data sets containing tweets about product reviews and tweets about people's thoughts on public policy are selected for experimentation. In the experiments done, it has been found that the SVM classifier using TFIDF and HFE shows better performance as compared to other classifiers. Using the feature sets, 97% accuracy and 98% F1- score is achieved in the aspect category prediction task.

Multilingual sentiment analysis during the pandemic using deep learning models

High mood swings and increased mortality with the COVID-19 outbreak, mood changes are a concern. COVID-19 is extremely contagious, and by December 20, 2021, the global death toll from COVID-19 exceeded 5.35 million. Under such circumstances, people changed their original way of life and isolated themselves from home to work online. Most people in this situation expressed their emotions through social media. In order to understand the changes of people's emotions during the pandemic, this research analyzed the emotions of people's messages posted on social media in different countries through the SenWave dataset by different deep learning models. The dataset contains 6 languages (English, Arabic, Spanish, French, Italian and Chinese) with more than 70,000 tweets, in this paper, only tweets collected from Twitter are concerned, which contains 11 labels. The dataset will be preprocessed and transferred into numerical vector, then TextCNN, BERT, and BERT-TextCNN models are implemented for the above problems. BERT model reached 0.69 and 0.53 on precision and recall respectively on Twitter dataset.

Enhancing COVID‐19 misinformation detection through novel attention mechanisms in NLP

AbstractThe rapid evolution of electronic media in recent decades has exponentially amplified the propagation of fake news, resulting in widespread confusion and misunderstanding among the masses, especially concerning critical topics like the COVID‐19 pandemic. Consequently, detecting fake news on social media has emerged as a prominent area of research, attracting significant attention. This article introduces a novel cascaded group multi‐head attention (CGMHA) model for COVID‐19 fake news detection. Our research collected Twitter datasets with accurate and fake tweets in Urdu. The novel CGMHA model and depth‐wise convolution capture local and global contextual information by employing multiple attention heads in a cascaded fashion, enabling a comprehensive understanding of fake news. While achieving state‐of‐the‐art performance, we also highlight challenges such as language variations and misinformation nuances in the detection process, contributing to a more comprehensive understanding of the complexities involved in combatting fake news. Our proposed model surpasses the performance of state‐of‐the‐art models in classifying fake news and achieves accuracy, F1 score, precision, and recall of 0.98, 0.96, 0.95, and 0.95, respectively.

The Effect Of Optimizers On The Generalizability Additive Neural Attention For Customer Support Twitter Dataset In Chatbot Application

When optimizing the performance of neural network-based chatbots, determining the optimizer is one of the most important aspects. Optimizers primarily control the adjustment of model parameters such as weight and bias to minimize a loss function during training. Adaptive optimizers such as ADAM have become a standard choice and are widely used for their invariant parameter updates' magnitudes concerning gradient scale variations, but often pose generalization problems. Alternatively, Stochastic Gradient Descent (SGD) with Momentum and the extension of ADAM, the ADAMW, offers several advantages. This study aims to compare and examine the effects of these optimizers on the chatbot CST dataset. The effectiveness of each optimizer is evaluated based on its sparse-categorical loss during training and BLEU in the inference phase, utilizing a neural generative attention-based additive scoring function. Despite memory constraints that limited ADAMW to ten epochs, this optimizer showed promising results compared to configurations using early stopping techniques. SGD provided higher BLEU scores for generalization but was very time-consuming. The results highlight the importance of finding a balance between optimization performance and computational efficiency, positioning ADAMW as a promising alternative when training efficiency and generalization are primary concerns.

Sentiment Analysis of Finnish Twitter Discussions on COVID-19 During the Pandemic

With the outbreak of the COVID-19 pandemic, researchers have studied how people reacted on social media during the pandemic. Sentiment analysis has been leveraged to gain insight. However, much of the research conducted on both sentiment analysis and social media analysis of COVID-19 often focuses on widespread languages, such as English and Chinese. This is partly due to the scarcity of resources for natural language processing and sentiment analysis for morphologically complex and less prevalent languages such as Finnish. This paper aims to analyze sentiments on Twitter in the Finnish language during the COVID-19 pandemic. We manually annotate with sentiments a random sample of 1943 tweets about COVID-19 in Finnish. We use it to build binomial and multinomial logistic regression models with Lasso penalty by exploiting ngrams and two existing sentiment lexicons. We also build two similar models using an existing (pre-COVID-19) Twitter dataset for comparison. The best-performing model for the Finnish language is then used to determine the trends of positive, negative, and neutral opinions on a collection of tweets in Finnish extracted between April 21 and June 18, 2020. The best sentiment polarity prediction model for the Finnish language attain 0.785 AUC, 0.710 balanced accuracy, and 0.723 macro-averaged F1 for predicting positive and negative polarity (binomial classification), and 0.667 AUC, 0.607 balanced accuracy, and 0.475 F1 when adding neutral tweets (multinomial classification). On the other hand, the pre-COVID-19 model trained on the same number of tweets exhibits higher accuracy for the multinomial model (0.687 balanced accuracy, and 0.588 F1). We hypothesize that this loss of performance is due to the COVID-19 context that makes sentiment analysis of neutral tweets more difficult for the machine learning algorithm to predict. Running the model on all the extracted Finnish tweets, we observe a decrease in negativity and an increase in positivity over the observed time as the Finnish government lifts restrictions. Our results show that applying an existing general-purpose sentiment analyzer on tweets that are domain-specific, such as COVID-19, provides lower accuracy. More effort in the future needs to be invested in using and developing sentiment analysis tools tailored to their application domain when conducting large-scale social media analysis of specific medical issues, such as a global pandemic.

Bayesian Graph Local Extrema Convolution with Long-tail Strategy for Misinformation Detection

It has become a cardinal task to identify fake information (misinformation) on social media, because it has significantly harmed the government and the public. There are many spam bots maliciously retweeting misinformation. This study proposes an efficient model for detecting misinformation with self-supervised contrastive learning. A B ayesian graph L ocal extrema C onvolution (BLC) is first proposed to aggregate node features in the graph structure. The BLC approach considers unreliable relationships and uncertainties in the propagation structure, and the differences between nodes and neighboring nodes are emphasized in the attributes. Then, a new long-tail strategy for matching long-tail users with the global social network is advocated to avoid over-concentration on high-degree nodes in graph neural networks. Finally, the proposed model is experimentally evaluated with two public Twitter datasets and demonstrates that the proposed long-tail strategy significantly improves the effectiveness of existing graph-based methods in terms of detecting misinformation. The robustness of BLC has also been examined on three graph datasets and demonstrates that it consistently outperforms traditional algorithms when perturbed by 15% of a dataset.

Perceived experts are prevalent and influential within an antivaccine community on Twitter.

Perceived experts (i.e. medical professionals and biomedical scientists) are trusted sources of medical information who are especially effective at encouraging vaccine uptake. The role of perceived experts acting as potential antivaccine influencers has not been characterized systematically. We describe the prevalence and importance of antivaccine perceived experts by constructing a coengagement network of 7,720 accounts based on a Twitter data set containing over 4.2 million posts from April 2021. The coengagement network primarily broke into two large communities that differed in their stance toward COVID-19 vaccines, and misinformation was predominantly shared by the antivaccine community. Perceived experts had a sizable presence across the coengagement network, including within the antivaccine community where they were 9.8% of individual, English-language users. Perceived experts within the antivaccine community shared low-quality (misinformation) sources at similar rates and academic sources at higher rates compared to perceived nonexperts in that community. Perceived experts occupied important network positions as central antivaccine users and bridges between the antivaccine and provaccine communities. Using propensity score matching, we found that perceived expertise brought an influence boost, as perceived experts were significantly more likely to receive likes and retweets in both the antivaccine and provaccine communities. There was no significant difference in the magnitude of the influence boost for perceived experts between the two communities. Social media platforms, scientific communications, and biomedical organizations may focus on more systemic interventions to reduce the impact of perceived experts in spreading antivaccine misinformation.

Information Propagation Prediction Based on Spatial–Temporal Attention and Heterogeneous Graph Convolutional Networks

With the development of deep learning and other technologies, the research of information propagation prediction has also achieved important research achievements. However, the existing information diffusion studies either focus on the attention relationships of users or they predict the information according to the diffusion relationships of users, which makes the prediction results have certain limitations. Therefore, a prediction model has been proposed spatial–temporal attention heterogeneous graph convolutional networks (STAHGCNs). First, we use GCN to learn user influence relationships and user behavior relationships, and we propose a user representation fusion mechanism to learn the user characteristics. Second, to account for the dynamics of user behavior, a temporal attention mechanism strategy is used to encode time into the heterogeneous graph to obtain a more expressive user representation. Finally, the obtained user representation is input into the multihead attention mechanism for information propagation prediction. Experimental results performed on the Twitter, Douban, Digg, and Memetracker datasets have shown that the proposed STAHGCN model increased by 8.80% and 6.74% at hits@N and map@N, respectively, which are significantly better than the original latest DyHGCN model. The proposed STAHGCN model effectively integrates spatial factors, such as time factor, user influence, and behavior, which greatly improves the accuracy of information propagation prediction and has great significance for rumor monitoring and malicious account detection.

Disastrous Event and Sub-Event Detection From Microblog Posts Using Bi-Clustering Method

Social media has become a nondetachable part of our life, with the exponential growth of usage in the past decade. Social sites like Twitter, Facebook, Instagram, Flickr, Weibo, etc., with their millions of user base, apart from being a source of entertainment, has proven to be a very useful mean for public opinion generation, news propagation and information broadcasting by authorities. Social media data analysis has been a popular research area for the past few years. Detecting subevents from social media posts to identify an unusual event that requires special attention, especially in a disaster situation, is one of the key researches in this domain. In this article, we have proposed a novel biclustering-based subevent detection method from the Twitter dataset for retrospective analysis of disaster events. First, we have clustered the data matrix using spectral co-clustering. Then we identified subevents (words) and formulated a ranking framework to find the top-ranked subevents within the clusters. Finally, through statistical analysis, we have shown that the proposed framework works better than other existing subevent detection methods.

User-Specific Loyalty Measure and Prediction Using Deep Neural Network From Twitter Data

Analyzing the social media data posted by users reveals insights about the business such as customers’ sentiment, limitations, improvements in services, users’ expectations, customer loyalty, and satisfaction. Customer satisfaction and loyalty are measured using net promoter score (NPS), which is based on a survey and often lacks the background of customer context. Several recent methods, net brand reputation (NBR) and net sentiment score (NSS), overcome NPS limitations and utilize social media data to gauge customer loyalty. The NBR and NSS also have the limitation that they do not provide the user-level score. In social media, existing user has an influence on a new user, which cannot be estimated in NBR and NSS. The main contribution of this research is developing novel user-and product-level metrics, which measure the influence, satisfaction, and customer loyalty. User sentiment score (USS) measures customer loyalty, overall influence factor (OIF) computes influence factor, effective influence on user (EIU) gives the effective impact on a new user, and satisfaction score (SATS) estimates satisfaction using user tweets. Social data influence score (SDIS) is proposed to measure the combined impact of user satisfaction and influence. We prepare a Twitter dataset of five popular online learning platforms: Coursera, Udemy, Udacity, Khanacademy, and edX, and analyzed our proposed metrics on them. Prediction of USS into loyal and not loyal customers is carried out using Twitter data and dense neural networks (DNNs). The performance of DNN is compared with random forest classifier (RFC), eXtreme gradient boosting (XGB), and support vector classifier (SVC). DNN gave the highest classification of 98.62% on tenfold cross validation.

Reactions to macro-level shocks and re-examination of adaptation theory using Big Data.

Since 2020, the world has faced two unprecedented shocks: lockdowns (regulation) and the invasion of Ukraine (war). Although we realise the health and economic effects of these shocks, more research is needed on the effect on happiness and whether the type of shock plays a role. Therefore, in this paper, we determine whether these macro-level shocks affected happiness, how these effects differ, and how long it takes for happiness to adapt to previous levels. The latter will allow us to test whether adaptation theory holds at the macro level. We use a unique dataset of ten countries spanning the Northern and Southern hemispheres derived from tweets extracted in real-time per country. Applying Natural Language Processing, we obtain these tweets' underlying sentiment scores, after which we calculate a happiness score (Gross National Happiness) and derive daily time series data. Our Twitter dataset is combined with Oxford's COVID-19 Government Response Tracker data. Considering the results of the Difference-in-Differences and event studies jointly, we are confident that the shocks led to lower happiness levels, both with the lockdown and the invasion shock. We find that the effect size is significant and that the lockdown shock had a bigger effect than the invasion. Considering both types of shocks, the adaptation to previous happiness levels occurred within two to three weeks. Following our findings of similar behaviour in happiness to both types of shocks, the question of whether other types of shocks will have similar effects is posited. Regardless of the length of the adaptation period, understanding the effects of macro-level shocks on happiness is essential for policymakers, as happiness has a spillover effect on other variables such as production, safety and trust.

A Systematic Review of the Sarcasm Detection in the Twitter Dataset

Text is the most significant contributor to data generated on the Internet. Understanding a person's opinion is an essential part of natural language processing. However, people's views can be skewed and inaccurate if people use sarcasm when they post status updates, comment on blogs, and review products and movies. Sarcasm detection has gained an important role in social networking platforms because it can impact many applications such as sentimental analysis, opinion mining, and stance detection. Twitter is rapidly growing in volume, and its analysis presents significant challenges in detecting sarcasm. Our research work focuses on various methodologies available for detection of sarcasm. Various papers from recent years were collected and review was carried out. This paper discusses the literature on sarcasm detection under the category of datasets, in different pre-processing, feature extraction, feature selection, classification algorithms, and performance measures. This paper discusses the literature on sarcasm detection under the category of datasets, in different pre-processing, feature extraction, feature selection, classification algorithms, and performance measures. This work explores existing approaches, challenges, and future scopes for sarcasm detection in the Twitter dataset. This review bringsto light the analysis ofsarcasm identification in Twitter data and is intended to serve as a resource for researchers and practitioners interested in sarcasm detection and text classification.

The anatomy of conspiracy theorists: Unveiling traits using a comprehensive Twitter dataset

The discourse around conspiracy theories is currently thriving amidst the rampant misinformation in online environments. Research in this field has been focused on detecting conspiracy theories on social media, often relying on limited datasets. In this study, we present a novel methodology for constructing a Twitter dataset that encompasses accounts engaged in conspiracy-related activities throughout the year 2022. Our approach centers on data collection that is independent of specific conspiracy theories and information operations. Additionally, our dataset includes a control group comprising randomly selected users who can be fairly compared to the individuals involved in conspiracy activities. This comprehensive collection effort yielded a total of 15K accounts and 37M tweets extracted from their timelines. We conduct a comparative analysis of the two groups across three dimensions: topics, profiles, and behavioral characteristics. The results indicate that conspiracy and control users exhibit similarity in terms of their profile metadata characteristics. However, they diverge significantly in terms of behavior and activity, particularly regarding the discussed topics, the terminology used, and their stance on trending subjects. In addition, we find no significant disparity in the presence of bot users between the two groups. Finally, we develop a classifier to identify conspiracy users using features borrowed from bot, troll and linguistic literature. The results demonstrate a high accuracy level (with an F1 score of 0.94), enabling us to uncover the most discriminating features associated with conspiracy-related accounts.

COVID-19 Event Extraction from Twitter via Extractive Question Answering with Continuous Prompts.

As COVID-19 ravages the world, social media analytics could augment traditional surveys in assessing how the pandemic evolves and capturing consumer chatter that could help healthcare agencies in addressing it. This typically involves mining disclosure events that mention testing positive for the disease or discussions surrounding perceptions and beliefs in preventative or treatment options. The 2020 shared task on COVID-19 event extraction (conducted as part of the W-NUT workshop during the EMNLP conference) introduced a new Twitter dataset for benchmarking event extraction from COVID-19 tweets. In this paper, we cast the problem of event extraction as extractive question answering using recent advances in continuous prompting in language models. On the shared task test dataset, our approach leads to over 5% absolute micro-averaged F1-score improvement over prior best results, across all COVID-19 event slots. Our ablation study shows that continuous prompts have a major impact on the eventual performance.

A neural probabilistic bounded confidence model for opinion dynamics on social networks

Opinion dynamics (OD) models simulate the evolution of personal opinions using computer science, social dynamics, physics, and complexity science tools on social networks. However, current opinion dynamics models mainly utilize agent-based models to simulate user interactions but still need to leverage large-scale data fully. Social networks typically involve a vast number of users and information, making it difficult for models to capture the complex interactions in real social networks. Additionally, many models overlook the specific mechanisms of user interactions, such as how network individuals influence and interact with one another, resulting in inaccuracies in modeling individual opinion evolution. This paper proposes a novel approach that integrates the probabilistic bounded confidence model (PBC) with deep learning, i.e., Neural Probabilistic Bounded Confidence (NPBC), to tackle these challenges. The NPBC model adopts an attention network, a long short-term memory (LSTM) network, and a feedforward neural network to approximate the evolution of individual opinions within networks. It satisfies the data approximation and the users’ opinion update rule that conforms to the PBC model. This model enhances the understanding of opinion dynamics, helping to develop effective strategies for opinion guidance in social networks. Finally, the proposed NPBC model is evaluated on three synthetic and Twitter datasets. The results demonstrate that the proposed NPBC model performs better than the six baseline methods in predicting user opinions. Compared to baseline models, the model’s accuracy (ACC) and comprehensive evaluation (F1) scores improve by up to 18% and 7%, respectively.

Mobilization, self-expression or argument? A computational method for identifying language styles in political discussion on Twitter

PurposeThis study develops a computational method to investigate the predominant language styles in political discussions on Twitter and their connections with users' online characteristics.Design/methodology/approachThis study gathers a large Twitter dataset comprising political discussions across various topics from general users. It utilizes an unsupervised machine learning algorithm with pre-defined language features to detect language styles in political discussions on Twitter. Furthermore, it employs a multinomial model to explore the relationships between language styles and users' online characteristics.FindingsThrough the analysis of over 700,000 political tweets, this study identifies six language styles: mobilizing, self-expressive, argumentative, narrative, analytic and informational. Furthermore, by investigating the covariation between language styles and users' online characteristics, such as social connections, expressive desires and gender, this study reveals a preference for an informational style and an aversion to an argumentative style in political discussions. It also uncovers gender differences in language styles, with women being more likely to belong to the mobilizing group but less likely to belong to the analytic and informational groups.Practical implicationsThis study provides insights into the psychological mechanisms and social statuses of users who adopt particular language styles. It assists political communicators in understanding their audience and tailoring their language to suit specific contexts and communication objectives.Social implicationsThis study reveals gender differences in language styles, suggesting that women may have a heightened desire for social support in political discussions. It highlights that traditional gender disparities in politics might persist in online public spaces.Originality/valueThis study develops a computational methodology by combining cluster analysis with pre-defined linguistic features to categorize language styles. This approach integrates statistical algorithms with communication and linguistic theories, providing researchers with an unsupervised method for analyzing textual data. It focuses on detecting language styles rather than topics or themes in the text, complementing widely used text classification methods such as topic modeling. Additionally, this study explores the associations between language styles and the online characteristics of social media users in a political context.

Visualizing emoji usage in geo-social media across time, space, and topic

Social media is ubiquitous in the modern world and its use is ever-increasing. Similarly, the use of emojis within social media posts continues to surge. Geo-social media produces massive amounts of spatial data that can provide insights into users' thoughts and reactions across time and space. This research used emojis as an alternative to text-based social media analysis in order to avoid the common obstacles of natural language processing such as spelling mistakes, grammatical errors, slang, and sarcasm. Because emojis offer a non-verbal means to express thoughts and emotions, they provide additional context in comparison to purely text-based analysis. This facilitates cross-language studies. In this study, the spatial and temporal usage of emojis were visualized in order to detect relevant topics of discussion within a Twitter dataset that is not thematically pre-filtered. The dataset consists of Twitter posts that were geotagged within Europe during the year 2020. This research leveraged cartographic visualization techniques to detect spatial-temporal changes in emoji usage and to investigate the correlation of emoji usage with significant topics. The spatial and temporal developments of these topics and their respective emojis were visualized as a series of choropleth maps and map matrices. This geovisualization technique allowed for individual emojis to be independently analyzed and for specific spatial or temporal trends to be further investigated. Emoji usage was found to be spatially and temporally heterogeneous, and trends in emoji usage were found to correlate with topics including the COVID-19 pandemic, several political movements, and leisure activities.