Categorization Of Tweets Research Articles

Categorization of tweets for damages: infrastructure and human damage assessment using fine-tuned BERT model.

Identification of infrastructure and human damage assessment tweets is beneficial to disaster management organizations as well as victims during a disaster. Most of the prior works focused on the detection of informative/situational tweets, and infrastructure damage, only one focused on human damage. This study presents a novel approach for detecting damage assessment tweets involving infrastructure and human damages. We investigated the potential of the Bidirectional Encoder Representations from Transformer (BERT) model to learn universal contextualized representations targeting to demonstrate its effectiveness for binary and multi-class classification of disaster damage assessment tweets. The objective is to exploit a pre-trained BERT as a transfer learning mechanism after fine-tuning important hyper-parameters on the CrisisMMD dataset containing seven disasters. The effectiveness of fine-tuned BERT is compared with five benchmarks and nine comparable models by conducting exhaustive experiments. The findings show that the fine-tuned BERT outperformed all benchmarks and comparable models and achieved state-of-the-art performance by demonstrating up to 95.12% macro-f1-score, and 88% macro-f1-score for binary and multi-class classification. Specifically, the improvement in the classification of human damage is promising.

Analyzing and Automating Customer Service Queries on Twitter Using Robotic Process Automation

Several product companies have turned to social media to analyze customer satisfaction and provide customer service to disgruntled customers. Product twitter handles are flooded with tweets every day. Customer service professionals struggle to find and resolve complaints from numerous tweets, which results in high wait times for a response, huge costs for the company, and frustrated customers. Automating customer services involves the use of Artificial Intelligence (AI) and Natural Language Processing (NLP) to emulate customer service offered by professionals. Pre-processing and deriving insights from real data are difficult and several start-ups cannot afford funds to maintain a data analysis team. Existing technologies for chatbots and analysis can be improved with new machine learning models, training existing models more and ensuring that responses generated resemble human interactions. BERT-CNN-BiLSTM modules were integrated into a model for sentiment analysis on scraped tweets from twitter using TWINT that achieved an accuracy of 96%. For the labeling and categorization of tweets, the logistic regression linear model achieved the highest accuracy of 97% compared to the other 3 classification models. For the automated chatbot, a model trained with BERT and open AI-GPT achieved an accuracy of 78%.

Examining Tweet Content and Engagement of Users With Tweets About Hikikomori in Japanese: Mixed Methods Study of Social Withdrawal.

BackgroundHikikomori is a form of severe social withdrawal that is particularly prevalent in Japan. Social media posts offer insight into public perceptions of mental health conditions and may also inform strategies to identify, engage, and support hard-to-reach patient populations such as individuals affected by hikikomori.ObjectiveIn this study, we seek to identify the types of content on Twitter related to hikikomori in the Japanese language and to assess Twitter users’ engagement with that content.MethodsWe conducted a mixed methods analysis of a random sample of 4940 Japanese tweets from February to August 2018 using a hashtag (#hikikomori). Qualitative content analysis included examination of the text of each tweet, development of a codebook, and categorization of tweets into relevant codes. For quantitative analysis (n=4859 tweets), we used bivariate and multivariate logistic regression models, adjusted for multiple comparisons, and estimated the predicted probabilities of tweets receiving engagement (likes or retweets).ResultsOur content analysis identified 9 codes relevant to tweets about hikikomori: personal anecdotes, social support, marketing, advice, stigma, educational opportunities, refuge (ibasho), employment opportunities, and medicine and science. Tweets about personal anecdotes were the most common (present in 2747/4859, 56.53% of the tweets), followed by social support (902/4859, 18.56%) and marketing (624/4859, 12.84%). In the adjusted models, tweets coded as stigma had a lower predicted probability of likes (−33 percentage points, 95% CI −42 to −23 percentage points; P<.001) and retweets (−11 percentage points, 95% CI −18 to −4 percentage points; P<.001), personal anecdotes had a lower predicted probability of retweets (−8 percentage points, 95% CI −14 to −3 percentage points; P=.002), marketing had a lower predicted probability of likes (−13 percentage points, 95% CI −21 to −6 percentage points; P<.001), and social support had a higher predicted probability of retweets (+15 percentage points, 95% CI 6-24 percentage points; P=.001), compared with all tweets without each of these codes.ConclusionsJapanese tweets about hikikomori reflect a unique array of topics, many of which have not been identified in prior research and vary in their likelihood of receiving engagement. Tweets often contain personal stories of hikikomori, suggesting the potential to identify individuals with hikikomori through Twitter.

Overview of EXIST 2021: sEXism Identification in Social neTworks

The paper describes the organization, goals, and results of the sEXism Identification in Social neTworks (EXIST) challenge, a shared task proposed for the first time at IberLEF 2021. EXIST 2021 proposes two challenges: sexism identifi-cation and sexism categorization of tweets and gabs, both in Spanish and English. We have received a total of 70 runs for the sexism identification task and 61 for the sexism categorization challenge, submitted by 31 different teams from 11 countries. We present the dataset, the evaluation methodology, an overview of the proposed systems, and the results obtained. The final dataset consists of more than 11,000 annotated texts from two social networks (Twitter and Gab) and its development has been supervised and monitored by experts in gender issues.

Exploring academic libraries’ use of Twitter: a content analysis

Purpose – This research aims to analyze academic libraries’ Twitter content and present a categorization framework for the study of their tweets. Design/methodology/approach – The research adopted a statistical descriptive analysis in addition to a content analysis of the tweets. Consequently, many categories and subcategories were created to classify the tweets according to different aspects. A total of 17 academic library accounts were examined. Findings – The findings show that academic libraries used Twitter as a multifaceted tool. “News and announcements” received the highest score as the type of information most often posted on Twitter by libraries, followed by “library collections” and “library services”. The subcategories that received the highest scores were “library marketing and news”, “answers and referrals” and “books”. Academic libraries showed a penchant for posting links more often than other content. Other results show different patterns of communication and interaction between libraries and their Twitter followers. Practical implications – The categorization of tweets provides a framework for understanding how academic libraries use Twitter. Originality/value – As little research can be found in the library and information science literature on Twitter content analysis, this research is expected to contribute significantly. It will also support academic libraries and librarians in using Twitter more efficiently and effectively.

Searchable talk as discourse practice on the internet: The case of “#bindersfullofwomen”

This paper examines the phrase #bindersfullofwomen as an example of searchable talk in the microblogging sphere of Twitter. The data were collected and sampled in terms of its temporal development on Twitter. Drawing on the content analysis combined with descriptive statistics, the paper identified and compared the Twitter users as Journalist users and Ordinary users who are engaged with Twitter communication. It shows that Journalists tend to initiate the topic at the early stage of communication. Soon afterwards, the Ordinary participants dominate the Twitter communication. The tweets are textually characterized into three different categories on the formal grounds: Hashtags, Hashtags plus texts and anything with a link" or "a retweet", in the heteroglossic context of micro-blogging sphere. The categorization of tweets contributes as a methodology to distinguish the discourse practice performed by different Twitter users. The results show that both Journalist and Ordinary favor to post the informative tweets throughout the sampled period. The quantity of interactive tweets is becoming less favorable; however, in contrast, the tweets on the indeterminate plane gain high popularity from the early phase of the period. Our comparison indicates that the boundary between Journalist and Ordinary seems to be indistinct in terms of their contents.

Differences in personal and professional tweets of scholars

Purpose: This study shows that there were differences in the use of Twitter by professors at universities in the Association of American Universities (AAU). Affordance use differed between the personal and professional tweets of professors. Framing behaviors were described that could impact the interpretation of tweets by audience members. Design/methodology/approach: A three phase research design was used that included surveys of professors, categorization of tweets by Amazon's Mechanical Turk workers (i.e., turkers), and categorization of tweets by active professors on Twitter. Findings: There were significant differences found between professors that reported having a Twitter account, significant differences found between types of Twitter accounts (personal, professional, or both), and significant differences in the affordances used in personal and professional tweets. Framing behaviors were described that may assist altmetric researchers in distinguishing between personal and professional tweets. Research limitations/implications (if applicable): The study is limited by the sample population, survey instrument, low survey response rate, and low Cohen's kappa. Practical implications (if applicable): An overview of various affordances found in Twitter is provided and a novel use of Amazon's Mechanical Turk for the categorization of tweets is described that can be applied to future altmetric studies. Originality/value: This work utilizes a socio-technical framework integrating social and psychological theories to interpret results from the tweeting behavior of professors and the interpretation of tweets by workers in Amazon's Mechanical Turk.

English

Twitter is one of the most popular micro- blogging website in today's globalized world. Twitter messages can be mined to gain valuable information. Although Twitter provides a list of most popular topics people tweet about known as Trending Topics in real time, it is often hard to understand what these trending topics are about. Therefore, various efforts are being made to classify these topics into general categories with high accuracy for better information retrieval. We propose the use of one of the classification algorithm called Naive Bayes for the categorization of tweets which has been discussed in this paper. It then proposes how the Map - Reduce paradigm can be applied to existing Naive Bayes algorithm to handle large number of tweets.

트위터 상의 이미지 이용에 관한 분석

Given the context that users are actively using social media with multimedia embedded information, the purpose of this study is to demonstrate how images are used within Twitter messages, especially in influential and favorited messages. In order to achieve the purpose of this study, the top 200 influential and favorited messages with images were selected out of 1,589 tweets related to “Boston bombing” in April 2013. The characteristics of the message, image use, and user are analyzed and compared. Two phases of the analysis were conducted on three data sets containing the top 200 influential messages, top 200 favorited messages, and general messages. In the first phase, coding schemes have been developed for conducting three categorical analyses: (1) categorization of tweets, (2) categorization of image use, and (3) categorization of users. The three data sets were then coded using the coding schemes. In the second phase, comparison analyses were conducted among influential, favorited, and general tweets in terms of tweet type, image use, and user. While messages expressing opinion were found to be most favorited, the messages that shared information were recognized as most influential to users. On the other hand, as only four image uses - information dissemination, illustration, emotive/persuasive, and information processing - were found in this data set, the primary image use is likely to be data-driven rather than object-driven. From the perspective of users, the user types such as government, celebrity, and photo-sharing sites were found to be favorited and influential. An improved understanding of how users’ image needs, in the context of social media, contribute to the body of knowledge of image needs. This study will also provide valuable insight into practical designs and implications of image retrieval systems or services.