Search results for: Tweet

Authors: Amira Dhokar, Lobna Hlaoua, Lotfi Ben Romdhane

Abstract:

Tweet contextualization (TC) is a new issue that aims to answer questions of the form 'What is this tweet about?' The idea of this task was imagined as an extension of a previous area called multi-document summarization (MDS), which consists in generating a summary from many sources. In both TC and MDS, the summary should ideally contain the most relevant information of the topic that is being discussed in the source texts (for MDS) and related to the query (for TC). Furthermore of being informative, a summary should be coherent, i.e. well written to be readable and grammatically compact. Hence, coherence is an essential characteristic in order to produce comprehensible texts. In this paper, we propose a new approach to improve readability and coherence for tweet contextualization based on bipartite graphs. The main idea of our proposed method is to reorder sentences in a given paragraph by combining most expressive words detection and HITS (Hyperlink-Induced Topic Search) algorithm to make up a coherent context.

Keywords: bipartite graphs, readability, summarization, tweet contextualization

Procedia PDF Downloads 161

Authors: Judith Enriquez-Gibson

Abstract:

This paper meddles with the notion of phatic communion introduced 90 years ago by Malinowski, who was a Polish-born British anthropologist. It explores the phatic in Twitter within the contents of tweets related to moocs (massive online open courses) as a topic or trend. It is not about moocs though. It is about practices that could easily be hidden or neglected if we let big or massive topics take the lead or if we simply follow the computational or secret codes behind Twitter itself and third party software analytics. It draws from media and cultural studies. Though at first it appears data-driven as I submitted data collection and analytics into the hands of a third party software, Twitonomy, the aim is to follow how phatic communion might be practised in a social media site, such as Twitter. Lurking becomes its research method to analyse mooc-related tweets. A total of 3,000 tweets were collected on 11 October 2013 (UK timezone). The emphasis of lurking is to engage with Twitter as a system of connectivity. One interesting finding is that a click is in fact a phatic practice. A click breaks the silence. A click in one of the mooc website is actually a tweet. A tweet was posted on behalf of a user who simply chose to click without formulating the text and perhaps without knowing that it contains #mooc. Surely, this mechanism is not about reciprocity. To break the silence, users did not use words. They just clicked the ‘tweet button’ on a mooc website. A click performs and maintains connectivity – and Twitter as the medium in attendance in our everyday, available when needed to be of service. In conclusion, the phatic culture of breaking silence in Twitter does not have to submit to the power of code and analytics. It is a matter of human code.

Keywords: click, Twitter, phatic communion, social media data, mooc

Procedia PDF Downloads 380

Authors: Noor Farizah Ibrahim, Christopher Durugbo

Abstract:

This study aims to investigate how the Twittersphere reacted during the recent historical event of robotic landing on a comet. The news is about Philae, a robotic lander from European Space Agency (ESA), which successfully made the first-ever rendezvous and touchdown of its kind on a nucleus comet on November 12, 2014. In order to understand how Twitter is practically used in spreading messages on historical events, we conducted an analysis of one-week tweet feeds that contain the #CometLanding hashtag. We studied the trends of tweets, the diffusion of the information and the characteristics of the social network created. The results indicated that the use of Twitter as a platform enables online communities to engage and spread the historical event through social media network (e.g. tweets, retweets, mentions and replies). In addition, it was found that comprehensible and understandable hashtags could influence users to follow the same tweet stream compared to other laborious hashtags which were difficult to understand by users in online communities.

Keywords: diffusion of information, hashtag, social media, Twitter

Procedia PDF Downloads 294

Authors: Zuzana Komrsková

Abstract:

This paper shows the connection between emoticons and politeness in written computer-mediated communication. It studies if there are some differences in the use of emoticon between Czech and English written tweets. My assumptions about the use of emoticons were based on the use of greetings and thanks in real, face to face situations. The first assumption, that welcome greeting phrase would be accompanied by positive emoticon was correct. But for the farewell greeting both positive and negative emoticons are possible. My results show lower frequency of negative emoticons in this context. I also found quite often both positive and negative emoticon in the same tweet. The expression of gratitude is associated with positive emotions. The results show that emoticons accompany polite phrases of greeting and thanks very often both in Czech and English. The use of emoticons with studied polite phrases shows that emoticons have become an integral part of these phrases.

Keywords: Czech, emoticon, english, politeness, twitter

Procedia PDF Downloads 377

Authors: Melissa Vang, Raheyma Khan, Haihua Chen

Abstract:

The coronavirus disease 2019 (COVID‐19) pandemic has changed people's lives from all over the world. More people have turned to Twitter to engage online and discuss the COVID-19 vaccine. This study aims to present a text mining approach to identify people's attitudes towards the COVID-19 vaccine on Twitter. To achieve this purpose, we collected 54,268 COVID-19 vaccine tweets from September 01, 2020, to November 01, 2020, then the BERT model is used for the sentiment and topic analysis. The results show that people had more negative than positive attitudes about the vaccine, and countries with an increasing number of confirmed cases had a higher percentage of negative attitudes. Additionally, the topics discussed in positive and negative tweets are different. The tweet datasets can be helpful to information professionals to inform the public about vaccine-related informational resources. Our findings may have implications for understanding people's cognitions and feelings about the vaccine.

Keywords: BERT, COVID-19 vaccine, sentiment analysis, topic modeling

Procedia PDF Downloads 117

Authors: Najiba Ouled Omar, Azza Harbaoui, Henda Ben Ghezala

Abstract:

Current research practices sentiment analysis with a focus on social networks, DEfi Fouille de Texte (DEFT) (Text Mining Challenge) evaluation campaign focuses on opinion mining and sentiment analysis on social networks, especially social network Twitter. It aims to confront the systems produced by several teams from public and private research laboratories. DEFT offers participants the opportunity to work on regularly renewed themes and proposes to work on opinion mining in several editions. The purpose of this article is to scrutinize and analyze the works relating to opinions mining and sentiment analysis in the Twitter social network realized by DEFT. It examines the tasks proposed by the organizers of the challenge and the methods used by the participants.

Keywords: opinion mining, sentiment analysis, emotion, polarity, annotation, OSEE, figurative language, DEFT, Twitter, Tweet

Procedia PDF Downloads 108

Authors: Sidi Yang, Haiyi Zhang

Abstract:

Twitter is a microblogging platform, where millions of users daily share their attitudes, views, and opinions. Using a probabilistic Latent Dirichlet Allocation (LDA) topic model to discern the most popular topics in the Twitter data is an effective way to analyze a large set of tweets to find a set of topics in a computationally efficient manner. Sentiment analysis provides an effective method to show the emotions and sentiments found in each tweet and an efficient way to summarize the results in a manner that is clearly understood. The primary goal of this paper is to explore text mining, extract and analyze useful information from unstructured text using two approaches: LDA topic modelling and sentiment analysis by examining Twitter plain text data in English. These two methods allow people to dig data more effectively and efficiently. LDA topic model and sentiment analysis can also be applied to provide insight views in business and scientific fields.

Keywords: text mining, Twitter, topic model, sentiment analysis

Procedia PDF Downloads 149

Authors: Niddal Imam, Vassilios G. Vassilakis

Abstract:

After analysing messages posted by health-related spam campaigns in Twitter Arabic hashtags, we found that these campaigns use unique hijacked accounts (we call them adversarial hijacked accounts) as adversarial examples to fool deployed ML-based spam detectors. Existing ML-based models build a behaviour profile for each user to detect hijacked accounts. This approach is not applicable for detecting spam in Twitter hashtags since they are computationally expensive. Hence, we propose an adversary-aware ML-based detector, which includes a newly designed feature (avg posts) to improve the detection of spam tweets posted by the adversarial hijacked accounts at a tweet-level in trending hashtags. The proposed detector was designed considering three key points: robustness, adaptability, and interpretability. The new feature leverages the account’s temporal patterns (i.e., account age and number of posts). It is faster to compute compared to features discussed in the literature and improves the accuracy of detecting the identified hijacked accounts by 73%.

Keywords: Twitter spam detection, adversarial examples, evasion attack, adversarial concept drift, account hijacking, trending hashtag

Procedia PDF Downloads 34

Authors: Diana Codat

Abstract:

Tweets' analysis is part of text mining. Each document is a written text. It's possible to apply the usual text search techniques, in particular by switching to the bag-of-words representation. But the tweets induce peculiarities. Some may enrich the analysis. Thus, their length is calibrated (at least as far as public messages are concerned), special characters make it possible to identify authors (@) and themes (#), the tweet and retweet mechanisms make it possible to follow the diffusion of the information. Conversely, other characteristics may disrupt the analyzes. Because space is limited, authors often use abbreviations, emoticons to express feelings, and they do not pay much attention to spelling. All this creates noise that can complicate the task. The tweets carry a lot of potentially interesting information. Their exploitation is one of the main axes of the analysis of the social networks. We show how to access Twitter-related messages. We will initiate a study of the properties of the tweets, and we will follow up on the exploitation of the content of the messages. We will work under R with the package 'twitteR'. The study of tweets is a strong focus of analysis of social networks because Twitter has become an important vector of communication. This example shows that it is easy to initiate an analysis from data extracted directly online. The data preparation phase is of great importance.

Keywords: data mining, language R, social networks, Twitter

Procedia PDF Downloads 149

Authors: Smah Almotiri

Abstract:

One of the most common fields of natural language processing (NLP) is sentimental analysis. The inferred feeling in the text can be successfully mined for various events using sentiment analysis. Twitter is viewed as a reliable data point for sentimental analytics studies since people are using social media to receive and exchange different types of data on a broad scale during the COVID-19 epidemic. The processing of such data may aid in making critical decisions on how to keep the situation under control. The aim of this research is to look at how sentimental states differed in a single geographic region during the lockdown at two different times.1162 tweets were analyzed related to the COVID-19 pandemic lockdown using keywords hashtags (lockdown, COVID-19) for the first sample tweets were from March 23, 2020, until April 23, 2020, and the second sample for the following year was from March 1, 2020, until April 4, 2020. Natural language processing (NLP), which is a form of Artificial intelligence, was used for this research to calculate the sentiment value of all of the tweets by using AFINN Lexicon sentiment analysis method. The findings revealed that the sentimental condition in both different times during the region's lockdown was positive in the samples of this study, which are unique to the specific geographical area of New Zealand. This research suggests applying machine learning sentimental methods such as Crystal Feel and extending the size of the sample tweet by using multiple tweets over a longer period of time.

Keywords: sentiment analysis, Twitter analysis, lockdown, Covid-19, AFINN, NodeJS

Procedia PDF Downloads 153

Authors: Chaohong Pan

Abstract:

Through public diplomacy on social media, governments have attempted to influence foreign public opinion. What is the impact of digital public diplomacy? Public diplomacy research often relies on content analysis to study the strategies employed by communicators but has rarely examined its actual impact on the audience. In addition, we do not know if giving a communicator an explicit label, as Twitter does with “government account”, would change the effects of the messages. Can the government label reduce the percussiveness of public diplomacy messages by sending a warning signal? Using a 2 × 2 survey experiment, the present paper contributes to the study of public diplomacy by randomly exposing American participants to four types of tweets from Chinese diplomats. The stimulus materials vary in terms of the tweets’ content (“positive-china” vs. “negative-US) and Twitter government labels (with vs. without the labels). I found that positive tweets about China have a significant positive effect on Americans’ attitudes toward China, whereas negative tweets about the US have little effect on their opinions. Furthermore, positive-China tweets are effective only on China-related issues, which indicates that Chinese diplomats’ tweets have limited effects on shaping a foreign audience’s attitudes toward their own country. Lastly, I find that labels largely have no impact on a diplomatic tweet’s effect. These results contribute to our understanding of the effects of public diplomacy in the digital age.

Keywords: public diplomacy, china, foreign public opinion, twitter

Procedia PDF Downloads 147

Authors: Anas Alsuhaibani

Abstract:

Observing debate and discussion over social media has been found to be a promising tool to investigate different types of opinion. On 23 June 2016, Brexit voters in the UK decided to depart from the EU, with 51.9% voting to leave. On Twitter, there had been a massive debate in this context, and the hashtag Brexit was allocated as number six of the most tweeted hashtags across the globe in 2016. The study aimed to investigate the sentiment and themes expressed in a sample of tweets during a political event (Brexit) in 2016. A sentiment and thematic analysis was conducted on 1304 randomly selected tweets tagged with the hashtag Brexit in Twitter for the period from 10 June 2016 to 7 July 2016. The data were coded manually into two code frames, sentiment and thematic, and the reliability of coding was assessed for both codes. The sentiment analysis of the selected sample found that 45.63% of tweets conveyed negative emotions while there were only 10.43% conveyed positive emotions. It also surprisingly resulted that 29.37% were factual tweets, where the tweeter expressed no sentiment and the tweet conveyed a fact. For the thematic analysis, the economic theme dominated by 23.41%, and almost half of its discussion was related to business within the UK and the UK and global stock markets. The study reported that the current UK government and relation to campaign themes were the most negative themes. Both sentiment and thematic analyses found that tweets with more than one opinion or theme were rare, 8.29% and 6.13%, respectively.

Keywords: Brexit, political opinion mining, social media, twitter

Procedia PDF Downloads 170

Authors: Min Chen, Rubi Gupta

Abstract:

Understanding and predicting stock market movements is a challenging problem. It is believed stock markets are partially driven by public sentiments, which leads to numerous research efforts to predict stock market trend using public sentiments expressed on social media such as Twitter but with limited success. Recently a microblogging website StockTwits is becoming increasingly popular for users to share their discussions and sentiments about stocks and financial market. In this project, we analyze the text content of StockTwits tweets and extract financial sentiment using text featurization and machine learning algorithms. StockTwits tweets are first pre-processed using techniques including stopword removal, special character removal, and case normalization to remove noise. Features are extracted from these preprocessed tweets through text featurization process using bags of words, N-gram models, TF-IDF (term frequency-inverse document frequency), and latent semantic analysis. Machine learning models are then trained to classify the tweets' sentiment as positive (bullish) or negative (bearish). The correlation between the aggregated daily sentiment and daily stock price movement is then investigated using Pearson’s correlation coefficient. Finally, the sentiment information is applied together with time series stock data to predict stock price movement. The experiments on five companies (Apple, Amazon, General Electric, Microsoft, and Target) in a duration of nine months demonstrate the effectiveness of our study in improving the prediction accuracy.

Keywords: machine learning, sentiment analysis, stock price prediction, tweet processing

Procedia PDF Downloads 124

Authors: Marc Solé, Francesc Giné, Magda Valls, Nina Bijedic

Abstract:

What people say on social media has turned into a rich source of information to understand social behavior. Specifically, the growing use of Twitter social media for political communication has arisen high opportunities to know the opinion of large numbers of politically active individuals in real time and predict the global political tendencies of a specific country. It has led to an increasing body of research on this topic. The majority of these studies have been focused on polarized political contexts characterized by only two alternatives. Unlike them, this paper tackles the challenge of forecasting Spanish political trends, characterized by multiple political parties, by means of analyzing the Twitters Users political tendency. According to this, a new strategy, named Tweets Analysis Strategy (TAS), is proposed. This is based on analyzing the users tweets by means of discovering its sentiment (positive, negative or neutral) and classifying them according to the political party they support. From this individual political tendency, the global political prediction for each political party is calculated. In order to do this, two different strategies for analyzing the sentiment analysis are proposed: one is based on Positive and Negative words Matching (PNM) and the second one is based on a Neural Networks Strategy (NNS). The complete TAS strategy has been performed in a Big-Data environment. The experimental results presented in this paper reveal that NNS strategy performs much better than PNM strategy to analyze the tweet sentiment. In addition, this research analyzes the viability of the TAS strategy to obtain the global trend in a political context make up by multiple parties with an error lower than 23%.

Keywords: political tendency, prediction, sentiment analysis, Twitter

Procedia PDF Downloads 209

Authors: Sergio Andres Rojas, Julian Benavides Franco, Juan Tomas Sayago

Abstract:

An increasing amount of research demonstrates how market mood affects financial markets, but their primary goal is to demonstrate how Trump's tweets impacted US interest rate volatility. Following that lead, this work evaluates the effect that Trump's tweets had during his presidency on local and international stock markets, considering not just volatility but the direction of the movement. Three indexes for Trump's tweets were created relating his activity with movements in the S&P500 using natural language analysis and machine learning algorithms. The indexes consider Trump's tweet activity and the positive or negative market sentiment they might inspire. The first explores the relationship between tweets generating negative movements in the S&P500; the second explores positive movements, while the third explores the difference between up and down movements. A pseudo-investment strategy using the indexes produced statistically significant above-average abnormal returns. The findings also showed that the pseudo strategy generated a higher return in the local market if applied to intraday data. However, only a negative market sentiment caused this effect on daily data. These results suggest that the market reacted primarily to a negative idea reflected in the negative index. In the international market, it is not possible to identify a pervasive effect. A rolling window regression model was also performed. The result shows that the impact on the local and international markets is heterogeneous, time-changing, and differentiated for the market sentiment. However, the negative sentiment was more prone to have a significant correlation most of the time.

Keywords: market sentiment, Twitter market sentiment, machine learning, natural dialect analysis

Procedia PDF Downloads 37

Authors: Teklu Urgessa, Joong Seek Lee

Abstract:

Social media is becoming the primary choice of activists to make their voices heard. This fact is coupled by two main reasons. The first reason is the emergence web 2.0, which gave the users opportunity to become content creators than passive recipients. Secondly the control of the mainstream mass media outlets by the governments and individuals with their political and economic interests. This paper aimed at exploring twitter data of network actors talking about the marathon silver medalists on Rio2016, who showed solidarity with the Oromo protesters in Ethiopia on the marathon race finish line when he won silver. The aim is to discover important insight using social network analysis and mining. The hashtag #FeyisaLelisa was used for Twitter network search. The actors’ network was visualized and analyzed. It showed the central influencers during first 10 days in August, were international media outlets while it was changed to individual activist in September. The degree distribution of the network is scale free where the frequency of degrees decay by power low. Text mining was also used to arrive at meaningful themes from tweet corpus about the event selected for analysis. The semantic network indicated important clusters of concepts (15) that provided different insight regarding the why, who, where, how of the situation related to the event. The sentiments of the words in the tweets were also analyzed and indicated that 95% of the opinions in the tweets were either positive or neutral. Overall, the finding showed that Olympic stage protest of the marathoner brought the issue of Oromo protest to the global stage. The new research framework is proposed based for event-based social network analysis and mining based on the practical procedures followed in this research for event-based social media sense making.

Keywords: human rights, Olympics, social media, network analysis, social network ming

Procedia PDF Downloads 215

Authors: Andrea Ando

Abstract:

The development of the blockchain has been a major innovation in the technology field. It opened the door to the creation of novel cyberassets and currencies. In more recent times, the non-fungible tokens have started to be at the centre of media attention. Their popularity has been increasing since 2021, and they represent the latest in the world of distributed ledger technologies and cryptocurrencies. It seems more and more likely that NFTs will play a more important role in our online interactions. They are indeed increasingly taking part in the arts and technology sectors. Their impact on society and the market is still very difficult to define, but it is very likely that there will be a turning point in the world of digital assets. There are some examples of their peculiar behaviour and effect in our contemporary tech-market: the former CEO of the famous social media site Twitter sold an NFT of his first tweet for around £2,1 million ($2,5 million), or the National Basketball Association has created a platform to sale unique moment and memorabilia from the history of basketball through the non-fungible token technology. Their growth, as imaginable, paved the way for civil disputes, mostly regarding their position under the current intellectual property law in each jurisdiction. In April 2022, the High Court of England and Wales ruled in the OpenSea case that non-fungible tokens can be considered properties. The judge, indeed, concluded that the cryptoasset had all the indicia of property under common law (National Provincial Bank v. Ainsworth). The research has demonstrated that the ruling of the High Court is not providing enough answers to the dilemma of whether minting an NFT is a violation or not of intellectual property and/or property rights. Indeed, if, on the one hand, the technology follows the framework set by the case law (e.g., the 4 criteria of Ainsworth), on the other hand, the question that arises is what is effectively protected and owned by both the creator and the purchaser. Then the question that arises is whether a person has ownership of the cryptographed code, that it is indeed definable, identifiable, intangible, distinct, and has a degree of permanence, or what is attached to this block-chain, hence even a physical object or piece of art. Indeed, a simple code would not have any financial importance if it were not attached to something that is widely recognised as valuable. This was demonstrated first through the analysis of the expectations of intellectual property law. Then, after having laid the foundation, the paper examined the OpenSea case, and finally, it analysed whether the expectations were met or not.

Keywords: technology, technology law, digital law, cryptoassets, NFTs, NFT, property law, intellectual property law, copyright law

Procedia PDF Downloads 61

Authors: Elcin Timur Cakmak, Ayse Oguzlar

Abstract:

This study presents the results of the tweets sent by Twitter users on social media about the Russia-Ukraine war by bigram and trigram methods. On February 24, 2022, Russian President Vladimir Putin declared a military operation against Ukraine, and all eyes were turned to this war. Many people living in Russia and Ukraine reacted to this war and protested and also expressed their deep concern about this war as they felt the safety of their families and their futures were at stake. Most people, especially those living in Russia and Ukraine, express their views on the war in different ways. The most popular way to do this is through social media. Many people prefer to convey their feelings using Twitter, one of the most frequently used social media tools. Since the beginning of the war, it is seen that there have been thousands of tweets about the war from many countries of the world on Twitter. These tweets accumulated in data sources are extracted using various codes for analysis through Twitter API and analysed by Python programming language. The aim of the study is to find the word sequences in these tweets by the n-gram method, which is known for its widespread use in computational linguistics and natural language processing. The tweet language used in the study is English. The data set consists of the data obtained from Twitter between February 24, 2022, and April 24, 2022. The tweets obtained from Twitter using the #ukraine, #russia, #war, #putin, #zelensky hashtags together were captured as raw data, and the remaining tweets were included in the analysis stage after they were cleaned through the preprocessing stage. In the data analysis part, the sentiments are found to present what people send as a message about the war on Twitter. Regarding this, negative messages make up the majority of all the tweets as a ratio of %63,6. Furthermore, the most frequently used bigram and trigram word groups are found. Regarding the results, the most frequently used word groups are “he, is”, “I, do”, “I, am” for bigrams. Also, the most frequently used word groups are “I, do, not”, “I, am, not”, “I, can, not” for trigrams. In the machine learning phase, the accuracy of classifications is measured by Classification and Regression Trees (CART) and Naïve Bayes (NB) algorithms. The algorithms are used separately for bigrams and trigrams. We gained the highest accuracy and F-measure values by the NB algorithm and the highest precision and recall values by the CART algorithm for bigrams. On the other hand, the highest values for accuracy, precision, and F-measure values are achieved by the CART algorithm, and the highest value for the recall is gained by NB for trigrams.

Keywords: classification algorithms, machine learning, sentiment analysis, Twitter

Procedia PDF Downloads 50

Authors: Martina Katalin Szabó, István Üveges

Abstract:

In this paper, the usage of negative emotive words is examined on the basis of a large Hungarian twitter-database via NLP methods. The data is analysed from a gender point of view, as well as changes in language usage over time. The term negative emotive word refers to those words that, on their own, without context, have semantic content that can be associated with negative emotion, but in particular cases, they may function as intensifiers (e.g. rohadt jó ’damn good’) or a sentiment expression with positive polarity despite their negative prior polarity (e.g. brutális, ahogy ez a férfi rajzol ’it’s awesome (lit. brutal) how this guy draws’. Based on the findings of several authors, the same phenomenon can be found in other languages, so it is probably a language-independent feature. For the recent analysis, 67783 tweets were collected: 37818 tweets (19580 tweets written by females and 18238 tweets written by males) in 2016 and 48344 (18379 tweets written by females and 29965 tweets written by males) in 2021. The goal of the research was to make up two datasets comparable from the viewpoint of semantic changes, as well as from gender specificities. An exhaustive lexicon of Hungarian negative emotive intensifiers was also compiled (containing 214 words). After basic preprocessing steps, tweets were processed by ‘magyarlanc’, a toolkit is written in JAVA for the linguistic processing of Hungarian texts. Then, the frequency and collocation features of all these words in our corpus were automatically analyzed (via the analysis of parts-of-speech and sentiment values of the co-occurring words). Finally, the results of all four subcorpora were compared. Here some of the main outcomes of our analyses are provided: There are almost four times fewer cases in the male corpus compared to the female corpus when the negative emotive intensifier modified a negative polarity word in the tweet (e.g., damn bad). At the same time, male authors used these intensifiers more frequently, modifying a positive polarity or a neutral word (e.g., damn good and damn big). Results also pointed out that, in contrast to female authors, male authors used these words much more frequently as a positive polarity word as well (e.g., brutális, ahogy ez a férfi rajzol ’it’s awesome (lit. brutal) how this guy draws’). We also observed that male authors use significantly fewer types of emotive intensifiers than female authors, and the frequency proportion of the words is more balanced in the female corpus. As for changes in language usage over time, some notable differences in the frequency and collocation features of the words examined were identified: some of the words collocate with more positive words in the 2nd subcorpora than in the 1st, which points to the semantic change of these words over time.

Keywords: gender differences, negative emotive words, semantic changes over time, twitter

Procedia PDF Downloads 170

Authors: Sudha Subramani, Hua Wang

Abstract:

Domestic Violence (DV) against women is now recognized to be a serious and widespread problem worldwide. There is a growing concern that violence against women has a global public health impact, as well as a violation of human rights. From the existing statistical surveys, it is revealed that there exists a strong relationship between DV and health issues of women like bruising, lacerations, depression, anxiety, flashbacks, sleep disturbances, hyper-arousal, emotional distress, sexually transmitted diseases and so on. This social problem is still considered as behind the closed doors issue and stigmatized topic. Women conceal their sufferings from family and friends, as they experience a lack of trust in others, feelings of shame and embarrassment among the society. Hence, women survivors of DV experience some barriers in seeking the support of specialized services such as health care access, crisis support, and legal guidance. Fortunately, with the popularity of social media like Facebook and Twitter, people share their opinions and emotional feelings to seek the social and emotional support, for sympathetic encouragement, to show compassion and empathy among the public. Considering the DV, social media plays a predominant role in creating the awareness and promoting the support services to the public, as we live in the golden era of social media. The various professional people like the public health researchers, clinicians, psychologists, social workers, national family health organizations, lawyers, and victims or their family and friends share the unprecedentedly valuable information (personal opinions and experiences) in a single platform to improve the social welfare of the community. Though each tweet or post contains a less informational value, the consolidation of millions of messages can generate actionable knowledge and provide valuable insights about the public opinion in general. Hence, this paper reports on an exploratory analysis of the effectiveness of social media for unobtrusive assessment of attitudes and awareness towards DV. In this paper, mixed methods such as qualitative analysis and text mining approaches are used to understand the social media disclosures of DV through the lenses of opinion sharing, anonymity, and support seeking. The results of this study could be helpful to avoid the cost of wide scale surveys, while still maintaining appropriate research conditions is to leverage the abundance of data publicly available on the web. Also, this analysis with data enrichment and consolidation would be useful in assisting advocacy and national family health organizations to provide information about resources and support, raise awareness and counter common stigmatizing attitudes about DV.

Keywords: domestic violence, social media, social stigma and support, women health

Procedia PDF Downloads 259

Authors: Julia Ive, Lucia Specia

Abstract:

Understanding and detecting toxicity is an important problem to support safer human interactions online. Our work focuses on the important problem of contextual toxicity detection, where automated classifiers are tasked with determining whether a short textual segment (usually a sentence) is toxic within its conversational context. We use “toxicity” as an umbrella term to denote a number of variants commonly named in the literature, including hate, abuse, offence, among others. Detecting toxicity in context is a non-trivial problem and has been addressed by very few previous studies. These previous studies have analysed the influence of conversational context in human perception of toxicity in controlled experiments and concluded that humans rarely change their judgements in the presence of context. They have also evaluated contextual detection models based on state-of-the-art Deep Learning and Natural Language Processing (NLP) techniques. Counterintuitively, they reached the general conclusion that computational models tend to suffer performance degradation in the presence of context. We challenge these empirical observations by devising better contextual predictive models that also rely on NLP data augmentation techniques to create larger and better data. In our study, we start by further analysing the human perception of toxicity in conversational data (i.e., tweets), in the absence versus presence of context, in this case, previous tweets in the same conversational thread. We observed that the conclusions of previous work on human perception are mainly due to data issues: The contextual data available does not provide sufficient evidence that context is indeed important (even for humans). The data problem is common in current toxicity datasets: cases labelled as toxic are either obviously toxic (i.e., overt toxicity with swear, racist, etc. words), and thus context does is not needed for a decision, or are ambiguous, vague or unclear even in the presence of context; in addition, the data contains labeling inconsistencies. To address this problem, we propose to automatically generate contextual samples where toxicity is not obvious (i.e., covert cases) without context or where different contexts can lead to different toxicity judgements for the same tweet. We generate toxic and non-toxic utterances conditioned on the context or on target tweets using a range of techniques for controlled text generation(e.g., Generative Adversarial Networks and steering techniques). On the contextual detection models, we posit that their poor performance is due to limitations on both of the data they are trained on (same problems stated above) and the architectures they use, which are not able to leverage context in effective ways. To improve on that, we propose text classification architectures that take the hierarchy of conversational utterances into account. In experiments benchmarking ours against previous models on existing and automatically generated data, we show that both data and architectural choices are very important. Our model achieves substantial performance improvements as compared to the baselines that are non-contextual or contextual but agnostic of the conversation structure.

Keywords: contextual toxicity detection, data augmentation, hierarchical text classification models, natural language processing

Procedia PDF Downloads 134

Authors: Jorge Cortés, Alejandra Martínez, Carlos Pérez, Anaid Simón

Abstract:

The presence of government, news media, and general citizenry in social media allows considering interactions between them as a form of political communication (i.e. the public exchange of contradictory discourses about politics). Twitter’s asymmetrical following model (users can follow, mention or reply to other users that do not follow them) could foster alternative democratic practices and have an impact on Mexican political culture, which has been marked by a lack of direct communication channels between these actors. The research aim is to assess Twitter’s role in political communication practices through the analysis of interaction dynamics between government, news media, and citizens by extracting and visualizing data from Twitter’s API to observe general behavior patterns. The hypothesis is that regardless the fact that Twitter’s features enable direct and horizontal interactions between actors, users repeat traditional dynamics of interaction, without taking full advantage of the possibilities of this medium. Through an interdisciplinary team including Communication Strategies, Information Design, and Interaction Systems, the activity on Twitter generated by the controversy over the presence of Uber in Mexico City was analysed; an issue of public interest, involving aspects such as public opinion, economic interests and a legal dimension. This research includes techniques from social network analysis (SNA), a methodological approach focused on the comprehension of the relationships between actors through the visual representation and measurement of network characteristics. The analysis of the Uber event comprised data extraction, data categorization, corpus construction, corpus visualization and analysis. On the recovery stage TAGS, a Google Sheet template, was used to extract tweets that included the hashtags #UberSeQueda and #UberSeVa, posts containing the string Uber and tweets directed to @uber_mx. Using scripts written in Python, the data was filtered, discarding tweets with no interaction (replies, retweets or mentions) and locations outside of México. Considerations regarding bots and the omission of anecdotal posts were also taken into account. The utility of graphs to observe interactions of political communication in general was confirmed by the analysis of visualizations generated with programs such as Gephi and NodeXL. However, some aspects require improvements to obtain more useful visual representations for this type of research. For example, link¬crossings complicates following the direction of an interaction forcing users to manipulate the graph to see it clearly. It was concluded that some practices prevalent in political communication in Mexico are replicated in Twitter. Media actors tend to group together instead of interact with others. The political system tends to tweet as an advertising strategy rather than to generate dialogue. However, some actors were identified as bridges establishing communication between the three spheres, generating a more democratic exercise and taking advantage of Twitter’s possibilities. Although interactions in Twitter could become an alternative to political communication, this potential depends on the intentions of the participants and to what extent they are aiming for collaborative and direct communications. Further research is needed to get a deeper understanding on the political behavior of Twitter users and the possibilities of SNA for its analysis.

Keywords: interaction, political communication, social network analysis, Twitter

Procedia PDF Downloads 190