First Author\[1\]\*, Second Author2 (Double Blind Review: Please do not type or edit anything here until final-camera ready submissions) 1First affiliation, City and Country (Please do not type or edit anything here, our editors will do the work for you) 2Second affiliation, City and Country (Please do not type or edit anything here, our editors will do the work for you)
ARTICLE INFO ABSTRACT

Article history:

Received XX Month 2024

Revised XX Month 2024

Accepted XX Month 2024

Online first

Published 1 September 2024

 This study visualises the social media sentiment of perception towards autistic people through an interactive dashboard. Big data visualization is leveraged to enhance positive perceptions towards autistic people. The sentiment analysis includes extracting, preparing, and classifying the social media sentiment data to gain insights into perceptions towards autistic people. Consequently, a visually appealing and informative dashboard was successfully developed using Power BI to visualise the perception. Overall, this study is significant since it has the potential to raise public awareness of autistic people and help society better understand and support them.

Keywords:

Keyword1

Keyword2

Keyword3

Keyword4

Keyword5

Keyword6

DOI:

10.24191/jcrinn.v9i2
# background of study Sentiment analysis is a valuable tool for understanding public opinion and emotions on specific topics including autism. It uses natural language processing and machine learning to identify and classify sentiments in text data and gain insights into the polarity and topic of opinions. In the context of autism, sentiment analysis has been used to examine emotional content in online messages related to autism spectrum disorders, including identifying positive, negative, and neutral emotions expressed in the messages (Ko et al.,2023). Additionally, sentiment analysis has been deemed important in assisting autistic individuals in interpreting feelings and emotions in written content, highlighting its potential role in improving communication and understanding for individuals on the autism spectrum (Bergman & Humble, 2022). ## Social Media Sentiment Analysis Social media sentiment analysis assesses target consumers' attitudes and emotions toward a product, service, or brand by analyzing their conversations on social media platforms (Barnhart, 2022). By leveraging sentiment analysis on social media, companies can gain valuable insight into the feelings and opinions of their audience, allowing them to make data-driven decisions and improve their brand reputation (Newberry, 2022). This technique has been applied in various areas, including mental health, where it can be used to identify mental health issues and stigma on social media platforms (Göksel et al., 2023). For example, a study on Turkish Twitter messages related to autism spectrum disorder (ASD) used artificial intelligence methods to analyze the emotional content of these messages (Göksel et al., 2023). The study found that there were fewer messages containing neutral emotions related to ASD and suggested further research should be conducted on different social media platforms or using other sentiment analysis techniques (Göksel et al., 2023). In the context of mental health and autism, sentiment analysis can be used to identify and analyze the emotions and feelings expressed in social media posts, enabling a better understanding of the experiences and challenges faced by people with these conditions (Herdiansyah et al., 2023). This information can be valuable for companies, researchers and policy makers in developing targeted interventions and support systems for people with mental health problems and autism (Obagbuwa et al., 2023). ## Sentiment Analysis on Perception Towards Autistic People Autism sentiment analysis involves assessing the emotions and attitudes expressed in written communication related to autism. This analysis can be performed using various methods, including rule-based sentiment analysis, dictionary-based sentiment analysis, and machine learning techniques (Bergman & Humble, 2022). One study focused on conducting emotional analysis of Turkish Twitter messages on autism using both qualitative and artificial intelligence-based methods (Göksel et al., 2023). The study found that the word “autism” was used as an insult by some users and there were fewer messages with neutral emotions. Overall, sentiment analysis of autism-related content can provide insights into the emotions and attitudes that individuals and the public express toward autism. This information can be valuable in understanding the needs of autistic individuals and addressing any stigma or negative perceptions associated with autism. ## Enhancing Positive Perception Towards Autistic People Through Data Visualisation While there are no specific studies on improving positive perceptions toward autistic people through data visualization, research has shown that visual sensory experiences are important for autistic people (Parmar et al., 2021). Additionally, studies have shown that public perception of autism can be influenced by media, social media, and awareness campaigns (Forum on Promoting Children’s Cognitive et al., 2016). Therefore, data visualization can be a powerful tool to improve public understanding of autism and challenge myths and stereotypes. One possible approach to improving positive perceptions toward autistic people through data visualization is to create visualizations that highlight the strengths and unique abilities of autistic people. For example, one study found that people with autism have better visual processing skills that could be highlighted in data visualizations (Visual Skills Superior in Autism, Study Says, 2011). By highlighting the strengths and unique abilities of autistic people and challenging negative stereotypes, data visualizations can help create a more positive perception of autism in society. # social media sentiment analysis towards autistic people Scientists and researchers have developed highly efficient techniques to handle data which requires the sentiment analysis to process data and know its polarity to make the right decision. It includes five data processing steps: data collection, text preparation, sentiment detection, sentiment classification, and output presentation as shown in Fig. 1. ![A diagram of a structure Description automatically generated](6682c1bf1b14d_media/media/image1.PNG) Fig. 1. Sentiment analysis process Those steps were adopted in this study and the detailed implementation is discussed in the following section. ## Data Collection This study collects the data from the social media platform Tiktok by extracting the comments on the video related to autism using the scraper tool, Instant Data Scraper. The extracted data were exported into an Excel file, as depicted in Fig. 2. ![A screenshot of a computer Description automatically generated](6682c1bf1b14d_media/media/image2.png) Fig. 2. Data extraction using Instant Data Scraper ## Text Preparation Subsequently, the dataset was prepared by cleaning it from errors and inconsistencies to ensure its quality and reliability. The cleaning process involved handling missing values, renaming columns’ names, removing unnecessary column, adding new columns and replacing unnecessary symbols. Fig. 4 illustrates one of the crucial cleaning processes: handling missing values. Referring to the Fig. 3, 49 rows of the profilImage, 4 rows of the username and 350 rows of the comment columns had missing values after the checking process conducted using the isnull() function. Hence, all the affected rows were drop using the dropna() function resulting a dataset free from missing values. ![A screenshot of a computer Description automatically generated](6682c1bf1b14d_media/media/image3.png) Fig. 3. Data cleaning: Handling missing values This process was repeated several times on different video resulting multiple data sets. Thus, all the data sets were merged into one set using Python. Consequently, 10,495 rows of data were combined in a single Excel file, that was ready to be further analysed. ## Sentiment Detection and Classification ## Sentiment detection consists of examining phrases and sentences extracted from reviews and ideas. The phrases were then classified according to the polarity of the opinions they contained, such as positive or negative, good or bad, and like or dislike. Based on these ratings, comments are tagged with the appropriate sentiment category. This process includes tokenizing the text, feeding it into the sentiment analysis algorithm, and interpreting the model output to classify sentiment. In this study, a robustly optimized BERT retraining (roBERTa) approach presented by Liu et al. (2019) is used to analyse and categorise each comment as positive, neutral or negative based on its textual content. The classified data is then structured into a format suitable for visualization, with each comment tagged with its respective sentiment category. This classification makes it easier to visualize the sentiments of perception towards autistic individuals. ## Data Visualisation ## Finally, the classified data was visualised through an interactive and visually appealing dashboard. The dashboard was developed using Microsoft Power BI by importing the cleaned and sentiment-labelled dataset from the Cloudera Hive data warehouse. Various types of visualizations such as bar charts, line charts, and pie charts were considered to display the key metrics of the number of total comments, sentiment distribution over time, and the correlation between sentiment and engagement metrics such as likes. An interactive element such as filters and slicers are also built in to allow users to drill down into specific data segments and customize their view based on their needs. # visualisation of sentiments on perception towards autistic people # The dashboard is intended to provide users with a comprehensive, easy-to-use platform to explore the data, gain insights and understand public perception towards autistic people. As demonstrated in Fig 4., the dashboard provides an overview of perception towards autistic people by visualising the bar chart of the trends of comments to show notable spikes in engagement in specific periods. The overview also includes the key metrics, including the total number of users, the total number of likes, and total number of comments. This visualization helps understand fluctuations in public discourse and engagement over time. ![](6682c1bf1b14d_media/media/image4.png) Fig. 4. The overview of perception towards autistic people Subsequently, the detailed insight into the sentiment distribution of comments over time is visualised using the clustered bar chart as shown in Fig. 5. Using the color-coded bars, the category of sentiment was indicated by red colour for negative perception, yellow colour for neutral perception, while green colour for positive perception. ![A screenshot of a computer screen Description automatically generated](6682c1bf1b14d_media/media/image5.png) Fig. 5. The sentiment category of perception Another important visual element in the dashboard is a word cloud highlighting positive and negative keywords. The size of each word indicates the frequency of its use. As illustrated in Fig. 6., (left) words such as “thank”, “bless” “love”, “beautiful”, “strong”, and “patient” are among the most positive words, while words such as “hard”, “suffer” “struggle”, “severe”, “anxiety”, and “die” in Fig. 6., (right) are the negative words expressed in the comment on the video related to autism. ![A close-up of words Description automatically generated](6682c1bf1b14d_media/media/image6.png) ![A close-up of words Description automatically generated](6682c1bf1b14d_media/media/image7.png) Fig. 6. (left) positive keyword (right) negative keyword To conclude, by visualising the social media sentiment, a supportive and encouraging public perception towards autistic people could be enhanced. # Acknowledgements/Funding (DOUBLE-BLIND reviewing. Leave the section as is. Only include Acknowledgements text in the final submission paper) This section is compulsory. The following is an example of an acknowledgement statement: The authors would like to acknowledge the support of Universiti Teknologi Mara (UiTM), Cawangan Negeri Sembilan, Kampus Kuala Pilah and Faculty of Applied Sciences, Universiti Teknologi MARA, Shah Alam, Selangor, Malaysia for providing the facilities and financial support on this research. # Conflict of interest statement (DOUBLE-BLIND reviewing. Leave the section as is. Only include CONFLICT OF INTERESTS text in the final submission paper) # This section is compulsory. The following is an example of a conflict-of-interest statement: The authors agree that this research was conducted in the absence of any self-benefits, commercial or financial conflicts and declare the absence of conflicting interests with the funders. # Authors’ contributions ***(Double Blind Review: Leave this section blank until final camera-ready submission)*:** # References Ko, K.-C., Liu, S.-K., Wei, C.-P., Hsieh, J.-S., & Yang, R.-H. (2023). To See the Invisible: An Empirical Comparison of Methods for Text-Based Sentiment Analysis of Online Contents From People With Autism Spectrum Condition. *International Journal of Market Research*, *65*(4), 402–422. Bergman, S., & Humble, N. (2022). Sentiment analysis for emotional navigation in written communication: What support do autistic people need? *European Conference on the Impact of Artificial Intelligence and Robotics*, *4*(1), Article 1. Barnhart, B. (2022, June 15). *The importance of social media sentiment analysis (and how to conduct it)*. Sprout Social. Newberry, C. (2022, September 29). *Social Media Sentiment Analysis: Tools and Tips for 2023*. Social Media Marketing & Management Dashboard. Göksel, P., Oban, V., Dikec, G., & Usta, M. (2023). Qualitative and Artificial Intelligence-Based Sentiment Analysis of Turkish Twitter Messages Related to Autism Spectrum Disorders. *Cureus*, *15*, e38446. Herdiansyah, H., Roestam, R., Kuhon, R., & Santoso, A. S. (2023). Their post tell the truth: Detecting social media users mental health issues with sentiment analysis. *Procedia Computer Science*, *216*, 691–697. Obagbuwa, I. C., Danster, S., & Chibaya, O. C. (2023). Supervised machine learning models for depression sentiment analysis. *Frontiers in Artificial Intelligence*, *6*. https://www.frontiersin.org/articles/10.3389/frai.2023.1230649 Parmar, K. R., Porter, C. S., Dickinson, C. M., Pelham, J., Baimbridge, P., & Gowen, E. (2021). Visual Sensory Experiences From the Viewpoint of Autistic Adults. *Frontiers in Psychology*, *12*. Forum on Promoting Children’s Cognitive, A., Board on Children, Y., Education, D. of B. and S. S. and, Division, H. and M., & National Academies of Sciences, E. (2016). Media and Public Perceptions and Misperceptions of Children with Disabilities. In *Ensuring Quality and Accessible Care for Children with Disabilities and Complex Health and Educational Needs: Proceedings of a Workshop*. National Academies Press (US). *Visual skills superior in autism, study says*. (2011, June 6). Spectrum | Autism Research News. Liu, Y., Ott, M., Goyal, N., Du, J., Joshi, M., et. al. (2019). RoBERTa: A Robustly Optimized BERT Pretraining Approach. arXiv. https://doi.org/10.48550/arXiv.1907.11692

A picture containing text, clipart Description automatically generated

 © 2024 by the authors. Submitted for possible open access publication under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
1. \* Corresponding author. *E-mail address*: (Add the e-mail in the final camera-ready submission)