
Conventional methods of data sampling in agriculture are time consuming, labor intensive, destructive, subject to human error and affected by field conditions. Thus, remote sensing technologies such as unmanned aerial vehicles (UAVs) became widely used as an alternative for data collection. Nevertheless, the big data captured by the UAVs is challenging to interpret. Therefore, machine learning algorithms (MLs) are used to interpret this data. However, the operational efficiency of those MLs is yet to be improved due to different sources affecting their modeling certainty. Therefore, this study aims to review different sources affecting the accuracy of MLs regression and classification interventions in precision agriculture. In this regard, 109 articles were identified in the Scopus database. The search was restricted to articles written in English, published during 2013–2020, and used UAVs as in-field data collection tools and ML algorithms for data analysis and interpretation. This systematic review will be the point of review for researchers to recognize the possible sources affecting the certainty of regression and classification results associated with MLs use. The recognition of those sources points out areas for improvement of MLs performance in precision agriculture. In this review, the performance of MLs is still evaluated in general, which opens the road for further detailed research.
Citation: Radhwane Derraz, Farrah Melissa Muharam, Noraini Ahmad Jaafar. Uncertainty sources affecting operational efficiency of ML algorithms in UAV-based precision agriculture: A 2013–2020 systematic review[J]. AIMS Agriculture and Food, 2023, 8(2): 687-719. doi: 10.3934/agrfood.2023038
[1] | Onoriode Kesiena, Henry K Onyeaka, Setri Fugar, Alexis K Okoh, Annabelle Santos Volgman . The top 100 Twitter influencers in cardiology. AIMS Public Health, 2021, 8(4): 743-753. doi: 10.3934/publichealth.2021058 |
[2] | Ali S. Abed Al Sailawi, Mohammad Reza Kangavari . Utilizing AI for extracting insights on post WHO's COVID-19 vaccination declaration from X (Twitter) social network. AIMS Public Health, 2024, 11(2): 349-378. doi: 10.3934/publichealth.2024018 |
[3] | Sara FL Kirk, Mary Forhan, Joshua Yusuf, Ashly Chance, Kathleen Burke, Nicole Blinn, Stephanie Quirke, Ximena Ramos Salas, Angela Alberga, Shelly Russell-Mayhew . Mapping changes in the obesity stigma discourse through Obesity Canada: a content analysis. AIMS Public Health, 2022, 9(1): 41-52. doi: 10.3934/publichealth.2022004 |
[4] | Noriko Sudo, Nobuyo Tsuboyama-Kasaoka, Ikuko Shimada, Keiichi Sato, Akiko Kubo . Attitudes toward using “Simple simulator for calculating nutritional food stocks in preparation for large-scale disasters” among local governmental personnel and public health dietitians in Japan: An explanatory mixed methods study. AIMS Public Health, 2022, 9(4): 734-757. doi: 10.3934/publichealth.2022051 |
[5] | Waled Amen Mohammed Ahmed, Sara Boutros Shokai, Insaf Hassan Abduelkhair, Amira Yahia Boshra . Factors Affecting Utilization of Family Planning Services in a Post-Conflict Setting, South Sudan: A Qualitative Study. AIMS Public Health, 2015, 2(4): 655-666. doi: 10.3934/publichealth.2015.4.655 |
[6] | Roberto Modena, Federico Schena . External training load and performance recovery after small-sided games in soccer: Insights for return-to-play management. AIMS Public Health, 2024, 11(1): 315-329. doi: 10.3934/publichealth.2024016 |
[7] | Okan Bulut, Cheryl N. Poth . Rapid assessment of communication consistency: sentiment analysis of public health briefings during the COVID-19 pandemic. AIMS Public Health, 2022, 9(2): 293-306. doi: 10.3934/publichealth.2022020 |
[8] | Sulistyawati Sulistyawati, Liena Sofiana, Sholehatun Khairul Amala, Rokhmayanti Rokhmayanti, Fardhiasih Dwi Astuti, Desi Nurfita . Pneumonia a neglected disease: A mixed-method study on the case-finding program in Indonesia. AIMS Public Health, 2020, 7(1): 81-91. doi: 10.3934/publichealth.2020008 |
[9] | Miguel Alarcón-Rivera, Carolina Cornejo-Mella, Camila Cáceres-Aravena, Yeny Concha-Cisternas, Paz Fernández-Valero, Eduardo Guzmán-Muñoz . Relationship between appendicular muscular mass index and physical function in older people. AIMS Public Health, 2024, 11(1): 130-140. doi: 10.3934/publichealth.2024006 |
[10] | Sushant K Singh . COVID-19: A master stroke of Nature. AIMS Public Health, 2020, 7(2): 393-402. doi: 10.3934/publichealth.2020033 |
Conventional methods of data sampling in agriculture are time consuming, labor intensive, destructive, subject to human error and affected by field conditions. Thus, remote sensing technologies such as unmanned aerial vehicles (UAVs) became widely used as an alternative for data collection. Nevertheless, the big data captured by the UAVs is challenging to interpret. Therefore, machine learning algorithms (MLs) are used to interpret this data. However, the operational efficiency of those MLs is yet to be improved due to different sources affecting their modeling certainty. Therefore, this study aims to review different sources affecting the accuracy of MLs regression and classification interventions in precision agriculture. In this regard, 109 articles were identified in the Scopus database. The search was restricted to articles written in English, published during 2013–2020, and used UAVs as in-field data collection tools and ML algorithms for data analysis and interpretation. This systematic review will be the point of review for researchers to recognize the possible sources affecting the certainty of regression and classification results associated with MLs use. The recognition of those sources points out areas for improvement of MLs performance in precision agriculture. In this review, the performance of MLs is still evaluated in general, which opens the road for further detailed research.
The occurrence of human stampedes is a significant health-related hazard in mass gathering events. These incidents involve a group of people running in the same direction, leading to injuries and deaths due to trauma [1]. As a result, mass gathering events have become a public health concern, often associated with significant mass casualty incidents [2],[3]. According to the World Health Organization (WHO), mass gathering events involve the aggregation of individuals at a specific place and time for a particular purpose, whether planned or spontaneous. Such events tend to overwhelm the preparation and allocation of resources by the host community or country [4]. As the global population continues to grow, and migration increases, mass gatherings have become more frequent, heightening the risk of such hazards [5]. Apart from stampedes, mass gatherings are associated with various other health hazards, including water and sanitation-related diseases, non-communicable diseases, exacerbation of pre-existing health conditions, transmission of communicable diseases, mental health and psychological disorders, thermal disorders such as dehydration, accidents, trauma, crush injuries, terrorist attacks, and alcohol and substance abuse [2],[3]. While many of these hazards are challenging to predict, implementing preventive measures can mitigate their impact.
The mechanisms underlying stampedes present a topic for ongoing research and understanding [6],[7]. Presently, it is known that these incidents typically stem from a triggering event that evokes an emotional response from the public, leading to a crowd crush, resulting in injuries, fatalities, and disabilities [8]. While stampedes are often associated with events such as sports gatherings, music festivals, and political rallies, they can occur in a variety of contexts [6]. Interestingly, the risk and fatality rates of stampedes tend to be higher in less developed countries compared to developed ones [9],[10]. A study [6] examining data between 1980 and 2007 revealed 215 stampedes worldwide, resulting in 14078 injuries and 7069 deaths. Notably, 27.9% of these incidents occurred in South Asia, while 25.1% were in Africa. Despite this compelling evidence, empirical research into the nature of stampedes remains insufficient [5]. There is still much to explore and comprehend to enhance our knowledge and to further implement effective preventive measures.
On Saturday, October 1, 2022, a tragic stampede occurred at a soccer stadium in Indonesia, resulting in a reported devastating loss of at least 125 lives, alongside numerous individuals being severely injured [11]. This unfortunate event, along with the recurrence of stampedes during sporting events, particularly in developing countries, emphasizes the urgent need for significant attention from the public health sector. In recent times, the widespread availability and extensive usage of social media platforms have emerged as influential factors shaping and reflecting people's knowledge, attitudes, and perceptions, as well as influencing communication patterns and information exchange [12]. Leveraging social media data has become a valuable tool for researchers to gain insights into the scale of disasters, providing valuable information for the development of prevention and mitigation strategies. Social media offers advantages such as real-time updates, vast amounts of collected data, and easy accessibility and searchability of information [13],[14]. Empirical studies increasingly acknowledge the role of big data from social media in rapidly assessing disaster responses and evaluating damage [15]. Despite numerous studies utilizing natural language processing techniques to gain insights into natural disasters through social media data [16]–[23], there remains a noticeable scarcity of research examining public perceptions and attitudes towards stampedes using these platforms [14]. Filling this research gap is crucial to understanding public sentiments and reactions to such incidents, ultimately helping to improve disaster preparedness and response strategies.
Stampedes, epidemics, and natural disasters are all emergencies that can potentially lead to loss of life. However, several distinctions set stampedes apart from epidemics and natural disasters. Stampedes tend to occur suddenly and are often brought under control relatively quickly, as compared to the gradual development of epidemics and natural disasters [24]. Additionally, the contagion effects of stampedes are immediate and confined to the incident's immediate vicinity, whereas epidemics and natural disasters can spread geographically, thereby affecting a broader population and potentially leading to long-term emotional and behavioral changes.
Despite the increasing volume of textual data available, leveraging big data from social media for qualitative research presents challenges. Issues such as premature sampling, early text selection, and difficulties in systematically and reproducibly processing and analyzing large volumes of unstructured texts can arise [25],[26]. However, advancements in unsupervised machine learning techniques and computational text analytics, including sentiment analysis and topic modeling, offer potential solutions to overcome these challenges [27],[28]. By utilizing natural language processing (NLP) techniques, researchers can identify public perceptions, attitudes, and detect latent topics embedded within big data [26],[28]. These approaches open up new avenues to gain valuable insights and understanding from the vast amount of social media data available for research.
Social media data has proven to be a valuable resource for studying various disasters in previous research. For instance, one study focused on crowd detection following the 2014 Shanghai New Year's Eve stampede, utilizing data from Weibo, a Chinese microblogging website [20]. Additionally, researchers have leveraged social media big data to study hurricanes and other natural hazards [15]–[19]. Notably, a recent study by Amoudi et al. [14] emphasized the crucial role of social media in tracking stampedes and mitigating risks through the analysis of social media data. These studies, along with other disaster-related research [15],[29]–[31], collectively demonstrate the invaluable contribution of social media data in advancing the field of disaster research. The utilization of such data opens up new possibilities for gaining insights into disaster events, understanding public sentiments and reactions, and developing strategies for disaster preparedness and response.
Our study employed NLP techniques to analyze the public discourse surrounding the Indonesian soccer stadium stampede, using textual data from a prominent social media platform. Our primary objective was to identify critical aspects of the disaster that could inform the development of effective public health strategies for disaster preparedness and responses. To achieve this goal, we sought to answer the following research questions:
RQ1: What sentiments have been expressed on social media in response to the Indonesian soccer stadium stampede?
RQ2: What underlying themes emerged from the social media discourse following the Indonesian soccer stadium stampede?
The approach employed in this study involved the following steps: data collection, data cleaning and preprocessing and data analyses. All steps were conducted using R (version 4.2.1). Access to Tweets for this study was obtained after consenting to the Twitter user agreement. Data for this study were obtained from Twitter, a popular microblogging platform that has approximately more than 300 million active users monthly [20]. Twitter allows users to post user-generated content containing up to 280 characters [14]. We used Twitter's official Application Programming Interface (API) and the “search_tweets()” function contained in the “twitteR” package to retrieve a random sample of publicly available English-language tweets related to the Indonesian soccer stampede, which were posted from 1st–5th October 2022. However, retweets and non-English language tweets were excluded.
This study did not require institutional review board approval as we used publicly available de-identified data.
We performed data cleaning and preprocessing techniques to remove noise, which are usually present in unstructured texts such as in social media data. This involved the replacement of contractions and the removal of special characters (“&”, “@”, “$”, “#”), numbers, punctuations, username mentions, non-American Standard Code for Information Interchange (ASCII) characters from strings, weblinks, unicode, whitespaces, emojis, sentence breaks, duplicate tweets and stop words. Additionally, texts were converted to lower case and only tweets containing more than four words were retained. All words were tokenized into single words. To maintain the anonymity of Twitter users, we replaced users' account names with ‘@username’ when presenting phrases and quotes.
A sentiment analysis examines opinions, sentiments and emotions contained in the unstructured textural data [32]. The classification of texts into sentiments can be performed using either lexical methods, machine learning algorithms or a combination of both [32]. Each of these approaches can be performed at either the aspect, sentence or document level. In this study, we employed the lexicon-based approach to the sentiment analysis using the “syuzhet” library in R to identify prevalent sentiments and emotions expressed in the corpus. In terms of sentiment polarity, the nrc algorithm was applied to an emotion dictionary to score each tweet into positive, neutral and negative sentiments. On the other hand, an emotion analysis was performed based on the National Research Council (NRC) Word-Emotion Association Lexicon dictionary. The algorithm of the NRC lexicon is comprised of 14182 words and applies an emotion dictionary to categorize tweets based on Plutchik's wheel of emotions into eight emotional classifications (anger, anticipation, disgust, fear, joy, sadness, surprise and trust) [33],[34]. To do so, we used the get_sentiment() function in R. This function calculated sentiment scores for each word in the text using the NRC Word-Emotion Association Lexicon. Thereafter, the sum of the sentiment scores for each emotion was computed.
The size and unstructured nature of social media data makes manual classification of its content or functions into themes a challenging task [18]. Fortunately, this can be overcome using computational approaches. One such approach is the use of topic modeling. Topic modeling is an unsupervised machine learning approach based on Bayes' hierarchical model, which is used to identify latent topics within textual data by analyzing patterns among words and groups documents with similar patterns [18],[35]. While different approaches to topic modeling exist (e.g., Latent Dirichlet Analysis), in this study, we employed structural topic modeling (STM) with spectral initialization to identify latent themes in the Twitter discourse of the Indonesian soccer stampede. STM extends the framework of LDA and has the advantage of incorporating metadata as model covariates, which could provide further insight into how topic prevalence varies by these covariates [35]. To perform the STM, we used the R package “stm” [36] to (i) estimate the model with the optimal number of topics, (ii) visualize, label and interpret topics and (iii) cluster topics.
Given that the optimal number of topics (K) were unknown a priori, we estimated the number of topics using the “searchK()” to fit models beginning with 5 topics to a maximum of 50 topics, in increments of 5. Afterwards, each model was evaluated to select the best potential candidate topic model by examining the semantic coherence-exclusivity plot. To select the best model, we iteratively compared metrics from the model comprised of the semantic coherence, exclusivity, residuals and held-out likelihood. We selected the model which possessed a high semantic coherence, high held-out likelihood, high exclusivity, and low residual (i.e., high external validity and most semantically coherent and distinct topics) [35],[37].
After selecting the model with an optimal number of topics, we further examined each topic for the commonly occurring words based on the highest probability, frequency and exclusivity (FREX), score, and lift metrics and then subjectively assigned labels and interpreted each topic. The highest probability metric refers to words with the highest frequency of occurrence within each topic. On the other hand, FREX estimates the weights of words based on their frequency in a topic and by their degree of exclusiveness to that particular topic. Lift assigns higher weights to words which occur less frequently in other topics. The score is a weighting of words based on the logarithm of the frequency of a word in a specific topic divided by the logarithm of the frequency of the word in other topics.
In addition to selecting the best topic model, we examined the variation of topic prevalence by sentiment polarity and performed a correlation network analysis to group topics into distinct clusters to examine the correlation between the topics. Clustering categorizes topics into several groups in such a way that the intra-cluster similarity is greater than inter-cluster similarity. By doing so, input documents which share similar traits are segregated and assigned to clusters to provide an improved contextual comprehension.
A total of 46340 original tweets in the English language were retrieved, originating from 10391 unique accounts. After data preprocessing and the removal of duplicates, 16054 tweets posted by 9998 accounts remained for analysis. Among the accounts included in the analysis, the majority (n = 8066, 80.7%) only sent one tweet. On average, each account contributed 1.6 tweets (SD = 2.8). The number of followers for each account ranged from 0 to 6402124, with a median of 995. The collected tweets received a total of 279422 retweets, with an average of 17.4 retweets per tweet (SD = 306.9). Additionally, the tweets garnered a total of 962807 favorites, averaging 60 favorites per tweet (SD = 1005.7) from other users. Furthermore, 17.5% of the accounts included in the analysis were verified. The most commonly used platform for posting tweets was Twitter for Android (21.9%), followed by the Twitter Web App (19.6%) and Twitter for iPhone (16.4%). Out of the 16054 tweets analyzed, they contained a total of 301746 words and 767321 characters.
There was generally a negative sentiment valence towards the soccer stampede, with a mean sentiment value of -1.21 (SD = 1.04). The majority of tweets (87.8%, n = 14078) were classified as expressing negative sentiments towards the soccer stampede, while 8.2% (n = 1315) expressed a positive sentiment and 4.0% (n = 647) expressed a neutral sentiment. Of the emotions expressed in the corpus of tweets, fear was the most common emotion (29.3%) expressed in the tweets, with topics including the fear about the possible rise in death toll, dangers associated with playing football in Indonesia and the firing of tear gas by the police.
[Examples: “I just saw about the stadium stampede in Indonesia and once again im reminding yall that human stampedes are so fucking unbelievably dangerous and thats exactly why its not cool or funny to wave around a toy gun or shout fire or bomb cos you think it'd be hilarious, people die fr”
and
“The stadium turned into a smoke-filled battleground when police fired tear gas.”].
Sadness was the second most common emotion (19.1%) expressed and tweets included themes of rowdiness associated with football and disappointment with the stampede.
[Example: “Deepest sympathies and condolences to the families of victims of #Malang city stampede. The EU stands with Indonesia ID in this moment of great sadness @username.”]
and
“Watching the football situation that happened in indonesia on tiktok fyp with sang dewi song make me wanna cry so bad ”.
Anger was the next most prevalent emotion expressed in tweets and accounted for 15.9%.
[Examples: “Their team lost, so they got “angry”, stormed the field, and now 127 people are dead.”
and
“I'm both devastated and furious to hear that at least 174 people have died after teargas was fired at a football match in Indonesia. No one should die of “chaos, overcrowding, trampling and suffocation”]
The least common emotions found in tweets were joy (2.6%), surprise (5.5%) anticipation (6.6%).
In determining the model with the optimal number of topics, the best potential candidate topic models were estimated to between 15 and 20 topic models. To obtain a more parsimonious model, we reevaluated the topic models by specifying the number of topics starting with 10 and incrementally added one topic at a time, to a maximum of 25. After examining the exclusivity-semantic coherence plot and the output from the model diagnostic metrics, the model with 19 topics was selected. Next, we examined the prevalence of topics and compared the variation across positive and negative stampede-related discourses. Table 1 illustrates the resulting 19-topic structure characterizing the soccer stampede discourse.
The most prominent topic was “deaths and mortality” (Topic 10), which referred to deaths occurring as a result of the stampede. A common narrative within topic “I'm sorry to hear about the tragedy of football in Malang, Indonesia on October 1st. I couldn't help thinking why the police threw tear gas at the supporters who didn't come down to the field until it killed many people. You cops are murderers. I was so sad to hear this news. The second and third prominent topics focused on how families were impacted by the stampede (topic 11) and expression of condolences (topic 15), respectively. Less prominent topics within the stampede-related discourse were calls for ban and suspension of the league (topic 16), imposition of sanctions (topic 13), and the sack of the Indonesian police chief (topic 6).
We compared the estimated topic prevalence across tweets by sentiment polarity. Figure 1 shows the topic prevalence contrast between positive and negative sentiments. We observed “Riot” and “Death and mortality” to have a strong prevalence in the positive sentiments. Other topics which had a significant prevalence in the positive sentiment included “death toll”, “sack of police chief”, “stampede trigger” and “panic”. The topics “probe”, “condolence”, and “ban and suspension” had a strong prevalence in the negative sentiment.
Figure 2 presents a heatmap showing the correlations among topics listed in table 1. We observed that panic was positively correlated with the stampede trigger, while the imposition of sanctions had a substantial negative correlation with family impact.
Topic | Topic label | Proportion (%) | Top-10 words | FREX | Lift |
T10 | Deaths & Mortality | 15.1 | peopl, match, kill, game, polic, break, dozen, tonight, grow, washington | kill, peopl, break, match, grow, dozen, game | grow, kill, washington, break, peopl, dozen, tonight |
T11 | Family Impact | 8.7 | dead, fan, match, player, arm, cop, lose, wife, daughter, teenag | arm, dead, player, fan, wife, daughter, cop | arm, daughter, wife, player, teenag, dead, cop |
T15 | Condolences | 8.1 | famili, victim, happen, live, condol, lost, sad, incid, game, prayer | condol, friend, peac, love, happen, rip, live | friend, heartfelt, passion, rip, condol, deepest, extend |
T14 | Stampede Trigger | 7.6 | tear, gas, polic, fire, pitch, trigger, invad, fan, control, angri | trigger, invad, pitch, tear, respond, gas, quell | shatter, invad, respond, watchdog, trigger, quell, invas |
T9 | Riot | 5.9 | iot, injur, report, video, leav, cnn, sky, unrest, footballmatch, photo | riot, cnn, report, injur, sky, video, policesay | policesay, cnn, sky, footballmatch, riot, report, video |
T3 | FIFA President reacts | 5.8 | tragedi, sport, presid, fifa, histori, day, dark, hope, widodo, weekend | presid, dark, widodo, tragedi, joko, infantino, comprehens | comprehens, dark, expos, fix, gianni, infantino, joko |
T7 | Location | 5.3 | malang, kanjuruhan, arema, event, persebaya, clash, broke, affect, sadden, surabay | persebaya, arema, surabaya, sadden, kanjuruhan, malang, deepli | impun, surabaya, corner, persebaya, tactic, arema, hide |
T4 | Death Toll | 5.2 | death, toll, children, offici, rise, mourn, result, revis, mass, relat | toll, revis, rise, relat, reach, jump, offici | jump, ministri, toll, earlier, reach, revis, stadiumstamped |
T16 | Ban & Suspend | 4.6 | leagu, ban, support, suspend, erupt, countri, human, intern, associ, play | leagu, suspend, play, ban, fuck, bet, danger | hooligan, institut, play, activ, asap, bet, fuck |
T13 | Impose Sanctions | 4.5 | disast, world, worst, deadliest, fifa, cup, explain, pssi, host, sanction | worst, cup, world, sanction, react, disast, pressur | asham, sever, statut, cup, worst, amateur, pressur |
T6 | Police chief sack | 4.2 | olic, java, east, chief, offic, sunday, provinc, local, support, remov | chief, east, provinc, java, remov, local, offic | chief, law, provinc, remov, sack, east, capac |
T8 | Probe | 4.2 | time, investig, team, york, nyt, secur, probe, set, includ, independ | york, nyt, independ, lee, investig, muktita, sui | form, ian, lee, role, surviv, austin, brief |
T18 | Panic | 4.0 | polic, exit, trampl, caus, panic, left, gate, fire, field, suffoc | panic, escap, caus, rush, panick, soccermatch, exit | chaotic, escap, vermaut, fled, panick, disastr, drive |
T17 | Survivor experiences | 3.7 | die, chao, due, hospit, lock, contribut, horror, wit, recount, blame | lock, contribut, horror, recount, die, survivor, wit | contribut, door, lock, recount, survivor, horror, delay |
T5 | Crowd response | 3.3 | crush, crowd, respons, mount, anger, justic, punish, call, cri, left | respons, justic, crush, mount, cri, crowd, punish | cri, justic, perpetr, respons, demand, mount, move |
T2 | News headlines | 3.2 | news, bbc, stori, read, top, terribl, score, horribl, guardian, come | news, bbc, read, articl, score, headlin, quot | headlin, quot, bbc, articl, news, found, read |
T12 | Vigil | 2.6 | saturday, night, hundr, victim, silenc, minut, octob, held, vigil, hurt | silenc, minut, octob, vigil, memori, night, kick | : imag, tribut, uefa, candlelight, kick, memori, minut |
T1 | Event evolution | 2.3 | violenc, watch, author, unfold, reuter, post, inindonesia, stadiumcrush, footballstamped, fear | violenc, inindonesia, stadiumcrush, reuter, footballstamped, timelin, answer | answer, footballstamped, inindonesia, stadiumcrush, timelin, rivalri, reuter |
T19 | Sequence of events | 1.8 | match, death, occur, led, troubl, game, stand, polic, peopl, saturday | occur, match, troubl, led, death, stand, game | troubl, led, occur, death, stand, match, game |
Given the unpredictability and increasing likelihood of disasters occurring during mass gathering events, it is crucial to comprehend the public's reactions to such incidents. The extensive amount of communication and rapid exchange of information on social media platforms, particularly Twitter, offers a valuable opportunity to gain insights into early conversations and discourse surrounding such phenomena, including disasters. In this study, we utilized advanced machine learning techniques to analyze Twitter data, aiming to understand the public's response to the soccer stadium stampede and to uncover underlying themes within the discourse. The findings from this study serve as an initial exploratory investigation, providing preliminary evidence upon which more comprehensive studies can be based.
Our findings revealed that in the days following the Indonesian soccer stadium stampede, public perceptions were predominantly negative. Emotions expressed in tweets were characterized by fear, anger, and sadness. Through a thematic analysis, we identified 19 themes within the Twitter discourse, focusing on the circumstances surrounding the stampede, the Indonesian government's response, and the crowd reactions. The majority of these themes were associated with negative sentiments. The thematic analysis was further supported by examining the correlation among the topics. Overall, while it is not surprising that the soccer stampede was largely negatively perceived by the public, a notable observation from our analysis was the limited discussion of prehospital medical interventions in the Twitter discourse of the disaster. This finding is consistent with a previous study [20] that utilized Twitter data to assess the Shanghai 2014 stampede. The themes identified through topic modeling did not indicate that an emergency medical response was a prominent topic of discussion.
Regardless of their purpose, an essential aspect of mass gathering events is the prevention and management of health hazards. When it comes to events such as soccer matches, which attract a large number of spectators and participants spread across various sections of the stadium, planning and mitigation strategies are often overwhelmed, inefficient, and uncoordinated, especially in the event of stampedes [38]. Previous studies have highlighted differences in the attitudes and behaviors of football fans and spectators compared to those of other sports, indicating higher tendencies for aggression among football fans compared to rugby fans [39],[40]. Such a significant finding suggests that medical care planning and responses for mass gathering events should be tailored according to the specific type of event. A key aspect of this is the need for a comprehensive risk assessment, which should be conducted prior to mass gathering events to determine whether the risk of disaster is high or low. Factors such as the timing of the event, crowd behavior, spectator demographics, and geographic and environmental factors should be taken into consideration during this assessment process. For example, Thackway et al. [41] identified certain characteristics of mass gatherings that require enhanced public health planning and response. These include the duration of the gathering, the number of international visitors, and the geographical extent, which are important factors in determining public health preparedness. While it may not be feasible to implement the same level of effort in all parts of the world, especially in developing countries and contexts with limited resources, such efforts acknowledge the unique aspects and purpose of different mass gathering events, as well as their associated risks and the necessary level of disaster preparedness. Another study [42] examined the perspectives of rescue authorities regarding factors that should be considered in mass gathering preparation. The findings from this study categorized preparedness efforts into the pre-planning phase, factors in the emergency plan, and actions during the event. Taken together, the extent of public health actions should be guided by the outcome of a risk assessment, the availability of resources, and the coordination of responses by relevant agencies [41].
During disasters such as stampedes, it is vital to address the mental health needs of victims and survivors. Our study revealed that tweets expressed negative emotions, including fear, sadness, and anger. These psychological reactions can significantly impact the health and well-being of those affected. Therefore, prioritizing the recruitment and training of professionals who can address the mental health needs of disaster victims and survivors is crucial.
Our findings underscore the importance of preventing disasters during mass gathering events. This aligns with one of the core functions of public health, which involves assessing and monitoring the health of communities and populations at risk to identify health problems and priorities [13]. By contributing to the existing body of literature, which indicates a higher likelihood of stampedes occurring in developing countries, particularly in Asia [6], we emphasize the role of social media as a valuable surveillance tool for advancing disaster science and research. Examining a prevalent health-related hazard associated with sporting events, including the aggravating factors and severity of the damage, has important policy implications. Centralized communication and coordination between the government, local host communities, security agencies, and health agencies during planned mass gatherings is essential. A notable research gap is the limited focus on noncommunicable complications during mass gatherings, with much attention given to infectious diseases (especially during events like COVID-19) [2],[22]. Allocating funding from governments and funding agencies to study the noncommunicable consequences of disasters would provide robust evidence to inform disaster prevention and mitigation policies.
Lastly, our study highlights the increasing role of machine learning and artificial intelligence in predicting and preventing disasters. This underlines the potential for leveraging these technologies in future research and disaster management practices. Embracing these advancements can enhance our disaster preparedness and response capabilities, ultimately leading to more effective and proactive measures.
Our study had several limitations. First, it is important to note that social media users, including Twitter users, may not represent the entire Indonesian population, as individuals of different ages and socio-demographic backgrounds have varying presence on social media and across different platforms [13],[43]. However, despite this limitation, social media users can still reflect a diverse range of demographics and geographic locations, making the information obtained from these platforms relevant for policy and practice.
Second, the prevalence of infodemic, misinformation, and disinformation on social media raises concerns about the validity of the findings from our study. It is crucial to interpret our findings with caution, considering the potential influence of either false or misleading information circulating on social media platforms. Despite these challenges, our study still provides valuable insights into the nature of the disaster and the level of preparedness and response, both in terms of medical and non-medical aspects.
Finally, it is important to acknowledge that our study focused on Twitter comments posted in the English language. This approach may have overlooked perspectives expressed in Bahasa Indonesian, which is the official language of Indonesia. Consequently, our findings may not fully capture the relevant insights that could have emerged from analyzing tweets in the local language, limiting the interpretation of our results.
This study utilized advanced machine learning techniques to analyze Twitter data and gain insights into public reactions and discourse following a soccer stadium stampede in Indonesia. The findings revealed that the public perceptions and emotions expressed on Twitter were predominantly negative, as characterized by fear, anger, and sadness. A thematic analysis identified various ideas related to the circumstances of the stampede, government responses, and crowd reactions, most of which were associated with negative sentiments. Interestingly, the analysis did not indicate a significant focus on the critical aspect of prehospital medical interventions in the Twitter discourse of the disaster.
From a management perspective, the findings from this study contribute to the existing literature by emphasizing the need for tailored planning and mitigation strategies for different types of events, considering factors such as crowd behavior, demographics, and environmental conditions. In addition, there is a need for comprehensive risk assessments and enhanced public health planning, which are essential for effective disaster preparedness. Additionally, this study underscores the importance of considering mental health interventions for victims and survivors of mass gathering hazards. The negative emotions expressed on Twitter indicate the potential impact on the well-being of individuals exposed to such disasters. Special attention should be given to addressing the mental health needs of affected individuals by recruiting and training professionals in disaster response. Theoretically, our findings add to the existing body of literature by highlighting the higher likelihood of stampedes occurring in developing countries, particularly in Asia. Moreover, this study demonstrates the utility of social media as a surveillance tool for advancing disaster science and research. Central communication and coordination between the government, local host communities, security agencies, and health agencies are essential for effective disaster response.
Future research directions should include a focus on noncommunicable consequences of mass gatherings, as existing studies tend to predominantly address infectious diseases. Allocation of funding and resources to study noncommunicable complications would provide robust evidence to inform disaster prevention and mitigation policies. Additionally, the role of machine learning and artificial intelligence in predicting and preventing disasters should be further explored by researchers.
The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.
[1] | UN FAO (2019) Tracking Progress on Food and Agriculture-related SDG Indicators. Available from: http://www.fao.org/sdg-progress-report/en/#chapeau. |
[2] | United Nations, Department of Economic and Social Affairs, Population Division (2019) World population prospects 2019: Highlights. Available from: https://population.un.org/wpp/publications/files/wpp2019_highlights.pdf. |
[3] |
Delanka-Pedige HMK, Cheng X, Munasinghe-Arachchige SP, et al. (2020) Conventional vs. algal wastewater technologies: Reclamation of microbially safe water for agricultural reuse. Algal Res 51: 102022. https://doi.org/10.1016/j.algal.2020.102022 doi: 10.1016/j.algal.2020.102022
![]() |
[4] |
Singh R, Singh GS (2017) Traditional agriculture: a climate-smart approach for sustainable food production. Energ Ecol Environ 2: 296–316. https://doi.org/10.1007/s40974-017-0074-7 doi: 10.1007/s40974-017-0074-7
![]() |
[5] | Singh P, Pandey PC, Petropoulos GP, et al. (2020) Hyperspectral remote sensing in precision agriculture: present status, challenges, and future trends, In: Pandey PC, Srivastava PK, Balzter H, et al., (Eds.), Hyperspectral remote sensing: Theory and applications, Amsterdam: Elsevier. https://doi.org/10.1016/b978-0-08-102894-0.00009-7 |
[6] |
Liu S, Baret F, Andrieu B, et al. (2017) Estimation of wheat plant density at early stages using high resolution imagery. Front Plant Sci 8: 739. https://doi.org/10.3389/fpls.2017.00739 doi: 10.3389/fpls.2017.00739
![]() |
[7] |
Archontoulis SV, Miguez FE (2015) Nonlinear regression models and applications in agricultural research. Agron J 107: 786–798. https://doi.org/10.2134/agronj2012.0506 doi: 10.2134/agronj2012.0506
![]() |
[8] | Bishop CM (2006) Pattern recognition and machine learning. New York: Springer. |
[9] |
Tilly N, Aasen H, Bareth G (2015) Fusion of plant height and vegetation indices for the estimation of barley biomass. Remote Sens 7: 11449–11480. https://doi.org/10.3390/rs70911449 doi: 10.3390/rs70911449
![]() |
[10] |
Tilly N, Hoffmeister D, Cao Q, et al. (2013) Precise plant height monitoring and biomass estimation with Terrestrial Laser Scanning in paddy rice. ISPRS Ann Photogramm Remote Sens Spatial Inf Sci Ⅱ-5/W2: 295–300. https://doi.org/10.5194/isprsannals-Ⅱ-5-W2-295-2013 doi: 10.5194/isprsannals-Ⅱ-5-W2-295-2013
![]() |
[11] |
Boursianis AD, Papadopoulou MS, Diamantoulakis P, et al. (2020) Internet of Things (IoT) and Agricultural Unmanned Aerial Vehicles (UAVs) in smart farming: A comprehensive review. Internet Things 18: 100187. https://doi.org/10.1016/j.iot.2020.100187 doi: 10.1016/j.iot.2020.100187
![]() |
[12] |
Chlingaryan A, Sukkarieh S, Whelan B (2018) Machine learning approaches for crop yield prediction and nitrogen status estimation in precision agriculture: A review. Comput Electron Agric 151: 61–69. https://doi.org/10.1016/j.compag.2018.05.012 doi: 10.1016/j.compag.2018.05.012
![]() |
[13] |
Cust EE, Sweeting AJ, Ball K, et al. (2019) Machine and deep learning for sport-specific movement recognition: A systematic review of model development and performance. J Sports Sci 37: 568–600. https://doi.org/10.1080/02640414.2018.1521769 doi: 10.1080/02640414.2018.1521769
![]() |
[14] |
García R, Aguilar J, Toro M, et al. (2020) A systematic literature review on the use of machine learning in precision livestock farming. Comput Electron Agric 179: 105826. https://doi.org/10.1016/j.compag.2020.105826 doi: 10.1016/j.compag.2020.105826
![]() |
[15] |
Sharma R, Kamble SS, Gunasekaran A, et al. (2020) A systematic literature review on machine learning applications for sustainable agriculture supply chain performance. Comput Oper Res 119: 104926. https://doi.org/10.1016/j.cor.2020.104926 doi: 10.1016/j.cor.2020.104926
![]() |
[16] |
Colorado JD, Cera-Bornacelli N, Caldas JS, et al. (2020) Estimation of nitrogen in rice crops from UAV-captured images. Remote Sens 12: 3396. https://doi.org/10.3390/rs12203396 doi: 10.3390/rs12203396
![]() |
[17] |
She B, Yang YY, Zhao ZG, et al. (2020) Identification and mapping of soybean and maize crops based on Sentinel-2 data. Int J Agric Biol Eng 13: 171–182. https://doi.org/10.25165/j.ijabe.20201306.6183 doi: 10.25165/j.ijabe.20201306.6183
![]() |
[18] |
Calvario G, Sierra B, Alarcón TE, et al. (2017) A multi-disciplinary approach to remote sensing through low-cost UAVs. Sensors 17: 1411. https://doi.org/10.3390/s17061411 doi: 10.3390/s17061411
![]() |
[19] |
Viljanen N, Honkavaara E, Näsi R, et al. (2018) A novel machine learning method for estimating biomass of grass swards using a photogrammetric canopy height model, images and vegetation indices captured by a drone. Agriculture 8: 70. https://doi.org/10.3390/agriculture8050070 doi: 10.3390/agriculture8050070
![]() |
[20] |
Ndikumana E, Minh DHT, Thu DNH, et al. (2018) Estimation of rice height and biomass using multitemporal SAR Sentinel-1 for Camargue, Southern France. Remote Sens 10: 1394. https://doi.org/10.3390/rs10091394 doi: 10.3390/rs10091394
![]() |
[21] |
Xu JX, Ma J, Tang YN, et al. (2020) Estimation of sugarcane yield using a machine learning approach based on uav-lidar data. Remote Sens 12: 2823. https://doi.org/10.3390/rs12172823 doi: 10.3390/rs12172823
![]() |
[22] |
Parreiras TC, Lense GHE, Moreira RS, et al. (2020) Using unmanned aerial vehicle and machine learning algorithm to monitor leaf nitrogen in coffee. Coffee Sci 15: 151736. https://doi.org/10.25186/.v15i.1736 doi: 10.25186/.v15i.1736
![]() |
[23] |
Jiang J, Cai W, Zheng H, et al. (2019) Using digital cameras on an unmanned aerial vehicle to derive optimum color vegetation indices for leaf nitrogen concentration monitoring in winter wheat. Remote Sens 11: 2667. https://doi.org/10.3390/rs11222667 doi: 10.3390/rs11222667
![]() |
[24] |
Chen Z, Miao Y, Lu J, et al. (2019) In-season diagnosis of winter wheat nitrogen status in smallholder farmer fields across a village using unmanned aerial vehicle-based remote sensing. Agronomy 9: 619. https://doi.org/10.3390/agronomy9100619 doi: 10.3390/agronomy9100619
![]() |
[25] |
Wang X, Miao Y, Dong R, et al. (2021) Machine learning-based in-season nitrogen status diagnosis and side-dress nitrogen recommendation for corn. Eur J Agron 123: 126193. https://doi.org/10.1016/j.eja.2020.126193 doi: 10.1016/j.eja.2020.126193
![]() |
[26] |
Maimaitijiang M, Sagan V, Sidike P, et al. (2020) Crop monitoring using satellite/UAV data fusion and machine learning. Remote Sens 12: 1357. https://doi.org/10.3390/RS12091357 doi: 10.3390/RS12091357
![]() |
[27] |
Nevavuori P, Narra N, Linna P, et al. (2020) Crop yield prediction using multitemporal UAV data and spatio-temporal deep learning models. Remote Sens 12: 4000. https://doi.org/10.3390/rs12234000 doi: 10.3390/rs12234000
![]() |
[28] | Goemans MX (1994) Advanced algorithms. Cambridge: Massachusetts Inst Technol. |
[29] | Pandey D, Niwaria K, Chourasia B (2019) Machine learning algorithms: A review. IRJET 6: 916–922. |
[30] |
Ha JG, Moon H, Kwak JT, et al. (2017) Deep convolutional neural network for classifying Fusarium wilt of radish from unmanned aerial vehicles. J Appl Remote Sens 11: 042621. https://doi.org/10.1117/1.jrs.11.042621 doi: 10.1117/1.jrs.11.042621
![]() |
[31] |
Kerkech M, Hafiane A, Canals R (2020) VddNet : Vine disease detection network based on multispectral images and depth map. Remote Sens 12: 3305. https://doi.org/10.3390/rs12203305 doi: 10.3390/rs12203305
![]() |
[32] |
Kwak GH, Park NW (2019) Impact of texture information on crop classification with machine learning and UAV images. Appl Sci 9: 643. https://doi.org/10.3390/app9040643 doi: 10.3390/app9040643
![]() |
[33] |
Lin Z, Guo W (2020) Sorghum panicle detection and counting using unmanned aerial system images and deep learning. Front Plant Sci 11: 534853. https://doi.org/10.3389/fpls.2020.534853 doi: 10.3389/fpls.2020.534853
![]() |
[34] |
Yang Q, Liu M, Zhang Z, et al. (2019) Mapping plastic mulched farmland for high resolution images of unmanned aerial vehicle using deep semantic segmentation. Remote Sens 11: 2008. https://doi.org/10.3390/rs11172008 doi: 10.3390/rs11172008
![]() |
[35] |
Chew R, Rineer J, Beach R, et al. (2020) Deep neural networks and transfer learning for food crop identification in UAV images. Drones 4: 7. https://doi.org/10.3390/drones4010007 doi: 10.3390/drones4010007
![]() |
[36] |
Gomez Selvaraj M, Vergara A, Montenegro F, et al. (2020) Detection of banana plants and their major diseases through aerial images and machine learning methods: A case study in DR Congo and Republic of Benin. ISPRS J Photogramm Remote Sens 169: 110–124. https://doi.org/10.1016/j.isprsjprs.2020.08.025 doi: 10.1016/j.isprsjprs.2020.08.025
![]() |
[37] |
Grüner E, Wachendorf M, Astor T (2020) The potential of UAV-borne spectral and textural information for predicting aboveground biomass and N fixation in legume-grass mixtures. PLoS One 15: 0234703. https://doi.org/10.1371/journal.pone.0234703 doi: 10.1371/journal.pone.0234703
![]() |
[38] |
Lambert JPT, Childs DZ, Freckleton RP (2019) Testing the ability of unmanned aerial systems and machine learning to map weeds at subfield scales: A test with the weed Alopecurus myosuroides (Huds). Pest Manag Sci 75: 2283–2294. https://doi.org/10.1002/ps.5444 doi: 10.1002/ps.5444
![]() |
[39] |
Malambo L, Popescu S, Ku N, et al. (2019) A deep learning semantic segmentation-based approach for field-level sorghum panicle counting. Remote Sens 11: 2939. https://doi.org/10.3390/rs11242939 doi: 10.3390/rs11242939
![]() |
[40] |
Osco LP, Junior M, Paula A, et al. (2020) Leaf nitrogen concentration and plant height prediction for maize using UAV-based multispectral imagery and machine learning techniques. Remote Sens 12: 3273. https://doi.org/10.3390/rs12193237 doi: 10.3390/rs12193237
![]() |
[41] |
Trevisan R, Pérez O, Schmitz N, et al. (2020) High-throughput phenotyping of soybean maturity using time series UAV imagery and convolutional neural networks. Remote Sens 12: 3617. https://doi.org/10.3390/rs12213617 doi: 10.3390/rs12213617
![]() |
[42] |
Wang T, Thomasson JA, Yang C, et al. (2020) Automatic classification of cotton root rot disease based on UAV remote sensing. Remote Sens 12: 1310. https://doi.org/10.3390/RS12081310 doi: 10.3390/RS12081310
![]() |
[43] |
Zha H, Miao Y, Wang T, et al. (2020) Improving unmanned aerial vehicle remote sensing-based rice nitrogen nutrition index prediction with machine learning. Remote Sens 12: 215. https://https://doi.org/10.3390/rs12020215 doi: 10.3390/rs12020215
![]() |
[44] |
Abdulridha J, Ampatzidis Y, Qureshi J, et al. (2020) Laboratory and UAV-based identification and classification of tomato yellow leaf curl, bacterial spot, and target spot diseases in tomato utilizing hyperspectral imaging and machine learning. Remote Sens 12: 2732. https://doi.org/10.3390/RS12172732 doi: 10.3390/RS12172732
![]() |
[45] |
Cao Y, Jiang K, Wu J, et al. (2020) Inversion modeling of japonica rice canopy chlorophyll content with UAV hyperspectral remote sensing. PLoS One 15: e0238530. https://doi.org/10.1371/journal.pone.0238530 doi: 10.1371/journal.pone.0238530
![]() |
[46] |
de Castro AI, Torres-Sánchez J, Peña JM, et al. (2018) An automatic random forest-OBIA algorithm for early weed mapping between and within crop rows using UAV imagery. Remote Sens 10: 285. https://doi.org/10.3390/rs10020285 doi: 10.3390/rs10020285
![]() |
[47] |
Durfee N, Ochoa CG, Mata-Gonzalez R (2019) The use of low-altitude UAV imagery to assess western juniper density and canopy cover in treated and untreated stands. Forests 10: 296. https://doi.org/10.3390/f10040296 doi: 10.3390/f10040296
![]() |
[48] |
Jin X, Liu S, Baret F, et al. (2017) Estimates of plant density of wheat crops at emergence from very low altitude UAV imagery. Remote Sens Environ 198: 105–114. https://doi.org/10.1016/j.rse.2017.06.007 doi: 10.1016/j.rse.2017.06.007
![]() |
[49] |
Karami A, Crawford M, Delp EJ (2020) Automatic plant counting and location based on a few-shot learning technique. IEEE J Sel Top Appl Earth Obs Remote Sens 13: 5872–5886. https://doi.org/10.1109/JSTARS.2020.3025790 doi: 10.1109/JSTARS.2020.3025790
![]() |
[50] |
Khan Z, Rahimi-Eichi V, Haefele S, et al. (2018) Estimation of vegetation indices for high-throughput phenotyping of wheat using aerial imaging. Plant Methods 14: 20. https://doi.org/10.1186/s13007-018-0287-6 doi: 10.1186/s13007-018-0287-6
![]() |
[51] |
Mart J, Apolo-apolo OE, Pe M (2020) A cloud-based environment for generating yield estimation maps from apple orchards using UAV imagery and a deep learning technique. Front Plant Sci 11: 1086. https://doi.org/10.3389/fpls.2020.01086 doi: 10.3389/fpls.2020.01086
![]() |
[52] |
Moeckel T, Dayananda S, Nidamanuri RR, et al. (2018) Estimation of vegetable crop parameter by multi-temporal UAV-borne images. Remote Sens 10: 805. https://doi.org/10.3390/rs10050805 doi: 10.3390/rs10050805
![]() |
[53] |
Moghimi A, Yang C, Anderson JA (2020) Aerial hyperspectral imagery and deep neural networks for high-throughput yield phenotyping in wheat. Comput Electron Agric 172: 105299. https://doi.org/10.1016/j.compag.2020.105299 doi: 10.1016/j.compag.2020.105299
![]() |
[54] |
Sapkota B, Singh V, Neely C, et al. (2020) Detection of Italian ryegrass in wheat and prediction of competitive interactions using remote-sensing and machine-learning techniques. Remote Sens 12: 2977. https://doi.org/10.3390/RS12182977 doi: 10.3390/RS12182977
![]() |
[55] |
Su J, Liu C, Coombes M, et al. (2018) Wheat yellow rust monitoring by learning from multispectral UAV aerial imagery. Comput Electron Agric 155: 157–166. https://doi.org/10.1016/j.compag.2018.10.017 doi: 10.1016/j.compag.2018.10.017
![]() |
[56] |
Wang T, Thomasson JA, Isakeit T, et al. (2020) A plant-by-plant method to identify and treat cotton root rot based on UAV remote sensing. Remote Sens 12: 2453. https://doi.org/10.3390/RS12152453 doi: 10.3390/RS12152453
![]() |
[57] |
Colorado JD, Francisco C, Mendez D, et al. (2020) A novel NIR-image segmentation method for the precise estimation of above-ground biomass in rice crops. PLoS ONE 15: e0239591. https://doi.org/10.1371/journal.pone.0239591 doi: 10.1371/journal.pone.0239591
![]() |
[58] |
Kerkech M, Hafiane A, Canals R (2020) Vine disease detection in UAV multispectral images using optimized image registration and deep learning segmentation approach. Comput Electron Agric 174: 105446. https://doi.org/10.1016/j.compag.2020.105446 doi: 10.1016/j.compag.2020.105446
![]() |
[59] | Lillesand T, Kiefer RW, Chipman J (2015) Remote sensing and image interpretation, 7th Eds., Hoboken: John Wiley & Sons. https://doi.org/10.14358/pers.81.8.615 |
[60] |
El Bilali A, Taleb A (2020) Prediction of irrigation water quality parameters using machine learning models in a semi-arid environment. J Saudi Soc Agric Sci 19: 439–451. https://doi.org/10.1016/j.jssas.2020.08.001 doi: 10.1016/j.jssas.2020.08.001
![]() |
[61] |
Sadeghi-tehran P, Virlet N, Ampe EM, et al. (2019) DeepCount : In-Field automatic quantification of wheat spikes using simple linear iterative clustering and deep convolutional neural networks. Front Plant Sci 10: 1176. https://doi.org/10.3389/fpls.2019.01176 doi: 10.3389/fpls.2019.01176
![]() |
[62] |
Han L, Yang G, Dai H, et al. (2019) Modeling maize above-ground biomass based on machine learning approaches using UAV remote-sensing data. Plant Methods 15: 10. https://doi.org/10.1186/s13007-019-0394-z doi: 10.1186/s13007-019-0394-z
![]() |
[63] |
Näsi R, Viljanen N, Kaivosoja J, et al. (2018) Estimating biomass and nitrogen amount of barley and grass using UAV and aircraft based spectral and photogrammetric 3D features. Remote Sens 10: 1082. https://doi.org/10.3390/rs10071082 doi: 10.3390/rs10071082
![]() |
[64] |
Ballesteros R, Ortega JF, Hernández D, et al. (2015) Characterization of vitis vinifera l. Canopy using unmanned aerial vehicle-based remote sensing and photogrammetry techniques. Am J Enol Vitic 66: 120–129. https://doi.org/10.5344/ajev.2014.14070 doi: 10.5344/ajev.2014.14070
![]() |
[65] |
Zhou C, Ye H, Xu Z, et al. (2019) Estimating maize-leaf coverage in field conditions by applying a machine learning algorithm to UAV remote sensing images. Appl Sci 9: 2389. https://doi.org/10.3390/app9112389 doi: 10.3390/app9112389
![]() |
[66] |
Csillik O, Cherbini J, Johnson R, et al. (2018) Identification of citrus trees from unmanned aerial vehicle imagery using convolutional neural networks. Drones 2: 39. https://doi.org/10.3390/drones2040039 doi: 10.3390/drones2040039
![]() |
[67] |
Dyson J, Mancini A, Frontoni E, et al. (2019) Deep learning for soil and crop segmentation from remotely sensed data. Remote Sens 11: 1859. https://doi.org/10.3390/rs11161859 doi: 10.3390/rs11161859
![]() |
[68] |
Cao F, Liu F, Guo H, et al. (2018) Fast detection of sclerotinia sclerotiorum on oilseed rape leaves using low-altitude remote sensing technology. Sensors 18: 4464 https://doi.org/10.3390/s18124464 doi: 10.3390/s18124464
![]() |
[69] |
Kawamura K, Asai H, Yasuda T, et al. (2020) Discriminating crops/weeds in an upland rice field from UAV images with the SLIC-RF algorithm. Plant Prod Sci 24: 198–215. https://doi.org/10.1080/1343943X.2020.1829490 doi: 10.1080/1343943X.2020.1829490
![]() |
[70] | Kitchenham B, Charters S (2007) Guidelines for performing Systematic Literature Reviews in Software Engineering: Version 2.3. Available from: https://www.researchgate.net/publication/302924724_Guidelines_for_performing_Systematic_Literature_Reviews_in_Software_Engineering. |
[71] | Cumpston M, Chandler J (2019) Planning a cochrane review, In: Higgins JPT, Thomas J, Chandler J, et al., (Eds.), Cochrane handbook for systematic reviews of interventions, Chichester: John Wiley & Sons. |
[72] | Li T, Higgins JPT, Deeks JJ (2019) Collecting data, In: Higgins JPT, Thomas J, Chandler J, et al., (Eds.), Cochrane handbook for systematic reviews of interventions, Chichester: John Wiley & Sons. |
[73] |
Moghimi A, Pourreza A, Zuniga-ramirez G, et al. (2020) A novel machine learning approach to estimate grapevine leaf nitrogen concentration using aerial multispectral imagery. Remote Sens 12: 3515. https://doi.org/10.3390/rs12213515 doi: 10.3390/rs12213515
![]() |
[74] |
Zhu W, Sun Z, Peng J, et al. (2019) Estimating maize above-ground biomass using 3D point clouds of multi-source unmanned aerial vehicle data at multi-spatial scales. Remote Sens 11: 2678. https://doi.org/10.3390/rs11222678 doi: 10.3390/rs11222678
![]() |
[75] |
Lu N, Zhou J, Han Z, et al. (2019) Improved estimation of aboveground biomass in wheat from RGB imagery and point cloud data acquired with a low-cost unmanned aerial vehicle system. Plant Methods 15: 17. https://doi.org/10.1186/s13007-019-0402-3 doi: 10.1186/s13007-019-0402-3
![]() |
[76] |
Lee H, Wang J, Leblon B (2020) Using linear regression, random forests, and support vector machine with unmanned aerial vehicle multispectral images to predict canopy nitrogen weight in corn. Remote Sens 12: 2071. https://doi.org/10.3390/rs12132071 doi: 10.3390/rs12132071
![]() |
[77] |
Zhang X, Han L, Dong Y, et al. (2019) A deep learning-based approach for automated yellow rust disease detection from high-resolution hyperspectral UAV images. Remote Sens 11: 1554. https://doi.org/10.3390/rs11131554 doi: 10.3390/rs11131554
![]() |
[78] |
Valente J, Sari B, Kooistra L, et al. (2020) Automated crop plant counting from very high-resolution aerial imagery. Precis Agric 21: 1366–1384. https://doi.org/10.1007/s11119-020-09725-3 doi: 10.1007/s11119-020-09725-3
![]() |
[79] |
Zhang J, Xie T, Yang C, et al. (2020) Segmenting purple rapeseed leaves in the field from UAV RGB imagery using deep learning as an auxiliary means for nitrogen stress detection. Remote Sens 12: 1403. https://doi.org/10.3390/RS12091403 doi: 10.3390/RS12091403
![]() |
[80] |
Zhou C, Hongbao Y, Jun H, et al. (2019) Automated counting of rice panicle by applying deep learning model to images from unmanned aerial vehicle platform. Sensors 19: 3106. https://doi.org/doi:10.3390/s19143106 doi: 10.3390/s19143106
![]() |
[81] |
Zhou X, Kono Y, Win A, et al. (2021) Predicting within-field variability in grain yield and protein content of winter wheat using UAV-based multispectral imagery and machine learning approaches. Plant Prod Sci 24: 137–151. https://doi:10.1080/1343943X.2020.1819165 doi: 10.1080/1343943X.2020.1819165
![]() |
[82] |
Gomez Selvaraj M, Valderrama M, Guzman D, et al. (2020) Machine learning for high-throughput field phenotyping and image processing provides insight into the association of above and below-ground traits in cassava (Manihot esculenta Crantz). Plant Methods 16: 87. https://doi.org/10.1186/s13007-020-00625-1 doi: 10.1186/s13007-020-00625-1
![]() |
[83] |
Zhou J, Tian Y, Yuan C, et al. (2019) Improved UAV opium poppy detection using an updated YOLOV3 model. Sensors 19: 4851. https://doi.org/10.3390/s19224851 doi: 10.3390/s19224851
![]() |
[84] |
Bah MD, Hafiane A, Canals R (2020) CRowNet: Deep network for crop row detection in UAV images. IEEE Access 8: 5189–5200. https://doi:10.1109/ACCESS.2019.2960873 doi: 10.1109/ACCESS.2019.2960873
![]() |
[85] |
Ampatzidis Y, Partel V, Costa L (2020) Agroview: Cloud-based application to process, analyze and visualize UAV-collected data for precision agriculture applications utilizing artificial intelligence. Comput Electron Agric 174: 105457. https://doi.org/10.1016/j.compag.2020.105457 doi: 10.1016/j.compag.2020.105457
![]() |
[86] |
Zhang D, Zhou X, Zhang J, et al. (2018) Detection of rice sheath blight using an unmanned aerial system with high-resolution color and multispectral imaging. PLoS One 13: e0187470. https://doi.org/10.1371/journal.pone.0187470 doi: 10.1371/journal.pone.0187470
![]() |
[87] |
Revill A, Florence A, MacArthur A, et al. (2019) The value of Sentinel-2 spectral bands for the assessment of winter wheat growth and development. Remote Sens 11: 2050. https://doi.org/10.3390/rs11172050 doi: 10.3390/rs11172050
![]() |
[88] |
Bareth G, Juliane B, Nora T, et al. (2016) A Comparison of UAV- and TLS-derived plant height for crop monitoring : Using polygon grids for the analysis of crop surface models (CSMs). PFG 2: 85–94. https://doi.org/10.1127/pfg/2016/0289 doi: 10.1127/pfg/2016/0289
![]() |
[89] |
Guo Y, Wang H, Wu Z, et al. (2020) Modified red blue vegetation index for chlorophyll estimation and yield prediction of maize from visible images captured by UAV. Sensors 20: 5055. https://doi.org/10.3390/s20185055 doi: 10.3390/s20185055
![]() |
[90] |
Di Nisio A, Adamo F, Acciani G, et al. (2020) Fast detection of olive trees affected by xylella fastidiosa from uavs using multispectral imaging. Sensors 20: 4915. https://doi.org/10.3390/s20174915 doi: 10.3390/s20174915
![]() |
[91] |
Zhang C, Craine WA, McGee RJ, et al. (2020) Image-based phenotyping of flowering intensity in cool-season crops. Sensors 20: 1450. https://doi.org/10.3390/s20051450 doi: 10.3390/s20051450
![]() |
[92] |
Kattenborn T, Eichel J, Wiser S, et al. (2020) Convolutional Neural Networks accurately predict cover fractions of plant species and communities in Unmanned Aerial Vehicle imagery. Remote Sens Ecol Conserv 6: 472–486. https://doi.org/10.1002/rse2.146 doi: 10.1002/rse2.146
![]() |
[93] |
Yamamoto K, Togami T, Yamaguchi N (2017) Super-resolution of plant disease images for the acceleration of image-based phenotyping and vigor diagnosis in agriculture. Sensors 17: 2557. https://doi.org/10.3390/s17112557 doi: 10.3390/s17112557
![]() |
[94] |
Zhou J, Yungbluth D, Vong CN, et al. (2009) Estimation of the maturity date of soybean breeding lines using UAV-based multispectral imagery. Remote Sens 11: 2075. https://doi.org/10.3390/rs11182075 doi: 10.3390/rs11182075
![]() |
[95] |
Pérez-Ortiz M, Peña JM, Gutiérrez PA, et al. (2016) Selecting patterns and features for between- and within- crop-row weed mapping using UAV-imagery. Expert Syst Appl 47: 85–94. https://doi.org/10.1016/j.eswa.2015.10.043 doi: 10.1016/j.eswa.2015.10.043
![]() |
[96] | Kuhn M, Johnson K (2013) Applied predictive modeling, New York: Springer. https://doi.org/10.1007/978-1-4614-6849-3 |
[97] |
Li J, Oswald C, Graef GL, et al. (2020) Improving model robustness for soybean iron de fi ciency chlorosis rating by unsupervised pre-training on unmanned aircraft system derived images. Comput Electron Agric 175: 105557. https://doi.org/10.1016/j.compag.2020.105557 doi: 10.1016/j.compag.2020.105557
![]() |
[98] |
Lu H, Cao Z (2020) TasselNetV2 + : A fast implementation for high-throughput plant counting from high-resolution RGB imagery. Front Plant Sci 11: 541960. https://doi.org/10.3389/fpls.2020.541960 doi: 10.3389/fpls.2020.541960
![]() |
[99] |
Freeman D, Gupta S, Hudson Smith D, et al. (2019) Watson on the farm: Using cloud-based artificial intelligence to identify early indicators of water stress. Remote Sens 11: 2645. https://doi.org/10.3390/rs11222645 doi: 10.3390/rs11222645
![]() |
[100] |
Ma J, Li Y, Du K, et al. (2020) Segmenting ears of winter wheat at flowering stage using digital images and deep learning. Comput Electron Agric 168: 105159. https://doi.org/10.1016/j.compag.2019.105159 doi: 10.1016/j.compag.2019.105159
![]() |
[101] |
Yang Q, Shi L, Han J, et al. (2020) A near real-time deep learning approach for detecting rice phenology based on UAV images. Agric Forest Meteorol 287: 107938. https://doi.org/10.1016/j.agrformet.2020.107938 doi: 10.1016/j.agrformet.2020.107938
![]() |
[102] |
Dian Bah M, Hafiane A, Canals R (2018) Deep learning with unsupervised data labeling for weed detection in line crops in UAV images. Remote Sens 10: 1690. https://doi.org/10.3390/rs10111690 doi: 10.3390/rs10111690
![]() |
[103] |
Su J, Coombes M, Liu C, et al. (2020) Machine learning-based crop drought mapping system by UAV remote sensing RGB imagery. Unmanned Syst 8: 71–83. https://doi.org/10.1142/S2301385020500053 doi: 10.1142/S2301385020500053
![]() |
[104] |
Yang M, Boubin JG, Ping H, et al. (2020) Adaptive autonomous UAV scouting for rice lodging assessment using edge computing with deep learning EDANet. Comput Electron Agric 179: 105817. https://doi.org/10.1016/j.compag.2020.105817 doi: 10.1016/j.compag.2020.105817
![]() |
[105] |
Jiang Q, Fang S, Peng Y, et al. (2019) UAV-based biomass estimation for rice-combining spectral, TIN-based structural and meteorological features. Remote Sens 11: 890. https://doi.org/10.3390/RS11070890 doi: 10.3390/RS11070890
![]() |
[106] |
Duarte-Carvajalino JM, Alzate DF, Ramirez AA, et al. (2018) Evaluating late blight severity in potato crops using unmanned aerial vehicles and machine learning algorithms. Remote Sens 10: 1513. https://doi.org/10.3390/rs10101513 doi: 10.3390/rs10101513
![]() |
[107] |
Qi G, Zhao G, Xi X (2020) Soil salinity inversion of winter wheat areas based on satellite-unmanned aerial vehicle-ground collaborative system in coastal of the Yellow River Delta. Sensors 20: 6521. https://doi.org/10.3390/s20226521 doi: 10.3390/s20226521
![]() |
[108] |
Fu Z, Jiang J, Gao Y, et al. (2020) Wheat growth monitoring and yield estimation based on multi-rotor unmanned aerial vehicle. Remote Sens 12: 508. https://doi.org/10.3390/rs12030508 doi: 10.3390/rs12030508
![]() |
[109] |
Niu Y, Zhang L, Zhang H, et al. (2019) Estimating above-ground biomass of maize using features derived from UAV-based RGB imagery. Remote Sens 11: 1261. https://doi.org/10.3390/rs11111261 doi: 10.3390/rs11111261
![]() |
[110] |
Dash JP, Watt MS, Paul TS, et al. (2019) Early detection of invasive exotic trees using UAV and manned aircraft multispectral and LiDAR Data. Remote Sens 11: 1812. https://doi.org/10.3390/rs11151812 doi: 10.3390/rs11151812
![]() |
[111] |
Wiesner-hanks T, Wu H, Stewart E, et al. (2019) Millimeter-level plant disease detection from aerial photographs via deep learning and crowdsourced data. Front Plant Sci 10: 1550. https://doi.org/10.3389/fpls.2019.01550 doi: 10.3389/fpls.2019.01550
![]() |
[112] |
Ampatzidis Y, Partel V (2019) UAV-based high throughput phenotyping in citrus utilizing multispectral imaging and artificial intelligence. Remote Sens 11: 410. https://doi.org/10.3390/rs11040410 doi: 10.3390/rs11040410
![]() |
[113] |
Dang LM, Wang H, Li Y, et al. (2020) Fusarium wilt of radish detection using rgb and near infrared images from unmanned aerial vehicles. Remote Sens 12: 2863. https://doi.org/10.3390/rs12172863 doi: 10.3390/rs12172863
![]() |
[114] |
Poblete T, Ortega-Farías S, Ryu D (2018) Automatic coregistration algorithm to remove canopy shaded pixels in UAV-borne thermal images to improve the estimation of crop water stress index of a drip-irrigated cabernet sauvignon vineyard. Sensors 18: 397. https://doi.org/10.3390/s18020397 doi: 10.3390/s18020397
![]() |
[115] |
Feng L, Zhang Z, Ma Y, et al. (2020) Alfalfa yield prediction using UAV-based hyperspectral imagery and ensemble learning. Remote Sens 12: 2028. https://doi.org/10.3390/rs12122028 doi: 10.3390/rs12122028
![]() |
[116] |
Qi G, Zhao G, Xi X (2020) Soil salinity inversion of winter wheat areas based on satellite-unmanned aerial vehicle-ground collaborative system in coastal of the Yellow River Delta. Sensors 20: 6521. https://doi.org/10.3390/s20226521 doi: 10.3390/s20226521
![]() |
[117] |
Wen D, Tongyu X, Fenghua Y, et al. (2018) Measurement of nitrogen content in rice by inversion of hyperspectral reflectance data from an unmanned aerial vehicle. Cienc Rural 48: e20180008. https://doi.org/10.1590/0103-8478cr20180008 doi: 10.1590/0103-8478cr20180008
![]() |
[118] |
Neupane B, Horanont T, Hung ND (2019) Deep learning based banana plant detection and counting using high-resolution red-green-blue (RGB) images collected from unmanned aerial vehicle (UAV). PLoS One 14: e0223906. https://doi.org/10.1371/journal.pone.0223906 doi: 10.1371/journal.pone.0223906
![]() |
[119] |
Predrag R, Đordevic V, Milic S, et al. (2020) Prediction of soybean plant density using a machine learning model and vegetation indices extracted from rgb images taken with a UAV. Agronomy 10: 1108. https://doi.org/10.3390/agronomy10081108 doi: 10.3390/agronomy10081108
![]() |
[120] |
Ampatzidis Y, Partel V, Meyering B, et al. (2019) Citrus rootstock evaluation utilizing UAV-based remote sensing and artificial intelligence. Comput Electron Agric 164: 104900. https://doi.org/10.1016/j.compag.2019.104900 doi: 10.1016/j.compag.2019.104900
![]() |
[121] |
Oh S, Chang A, Ashapure A, et al. (2020) Plant counting of cotton from UAS imagery using deep learning-based object detection framework. Remote Sens 12: 2981. https://doi.org/doi.org/10.3390/rs12182981 doi: 10.3390/rs12182981
![]() |
[122] |
Revill A, Florence A, Macarthur A, et al. (2020) Quantifying uncertainty and bridging the scaling gap in the retrieval of leaf area index by coupling sentinel-2 and UAV observations. Remote Sens 12: 1843. https://doi.org/10.3390/rs12111843 doi: 10.3390/rs12111843
![]() |
[123] |
Xu J-X, Ma J, Tang YN, et al. (2020) Estimation of sugarcane yield using a machine learning approach based on uav-lidar data. Remote Sens 12: 2823. https://doi.org/10.3390/rs12172823 doi: 10.3390/rs12172823
![]() |
[124] |
Carl C, Lehmann JRK, Landgraf D, et al. (2019) Robinia pseudoacacia L. in short rotation coppice: Seed and stump shoot reproduction as well as UAS-based spreading analysis. Forests 10: 235. https://doi.org/10.3390/f10030235 doi: 10.3390/f10030235
![]() |
[125] |
Mazzia V, Comba L, Khaliq A, et al. (2020) UAV and machine learning based refinement of a satellite-driven vegetation index for precision agriculture. Sensors 20: 2530. https://doi.org/10.3390/s20092530 doi: 10.3390/s20092530
![]() |
[126] |
Yao L, Van De Zedde R, Kowalchuk G (2021) Recent developments and potential of robotics in plant eco-phenotyping. Emerg Top Life Sci 5: 289–300. https://doi.org/10.1042/ETLS20200275 doi: 10.1042/ETLS20200275
![]() |
[127] |
Lakhiar IA, Jianmin G, Syed TN, et al. (2018) Monitoring and control systems in agriculture using intelligent sensor techniques: A review of the aeroponic system. J Sensors 2018: 8672769. https://doi.org/10.1155/2018/8672769 doi: 10.1155/2018/8672769
![]() |
[128] | Hägg L, Sandberg J (2021) The Engineer's Guide to Tank Gauging. Available from: https://www.emerson.com/documents/automation/-engineer-s-guide-to-tank-gauging-en-175314.pdf. |
[129] |
Adão T, Hruška J, Pádua L, et al. (2017) Hyperspectral imaging: A review on UAV-based sensors, data processing and applications for agriculture and forestry. Remote Sens 9: 1110. https://doi.org/10.3390/rs9111110 doi: 10.3390/rs9111110
![]() |
[130] |
Atefi A, Ge Y, Pitla S, et al. (2021) Robotic technologies for high-throughput plant phenotyping: Contemporary reviews and future perspectives. Front Plant Sci 12: 611940. https://doi.org/10.3389/fpls.2021.611940 doi: 10.3389/fpls.2021.611940
![]() |
[131] | Sharifi M, Sevier S, Zhang H, et al. (2021) Mechatronics for a LiDAR-based mobile robotic platform for pasture biomass measurement. In: Billingsley J., Brett P. (Eds.), Mechatronics and machine vision in Practice 4. Cham: Springer, 71–80. https://doi.org/10.1007/978-3-030-43703-9_6 |
[132] |
Azeta J, Bolu CA, Alele F, et al. (2019) Application of mechatronics in agriculture: A review. J Phys Conf Ser 1378: 032006. https://doi.org/10.1088/1742-6596/1378/3/032006 doi: 10.1088/1742-6596/1378/3/032006
![]() |
[133] |
Sharma A, Vans E, Shigemizu D, et al. (2019) DeepInsight: A methodology to transform a non-image data to an image for convolution neural network architecture. Sci Rep 9: 11399. https://doi.org/10.1038/s41598-019-47765-6 doi: 10.1038/s41598-019-47765-6
![]() |
[134] |
Zhu Y, Brettin T, Xia F, et al. (2021) Converting tabular data into images for deep learning with convolutional neural networks. Sci Rep 11: 11325. https://doi.org/10.1038/s41598-021-90923-y doi: 10.1038/s41598-021-90923-y
![]() |
[135] | Buturović L, Miljković D (2020) A novel method for classification of tabular data using convolutional neural networks. BioRxiv. https://doi.org/10.1101/2020.05.02.074203 |
Topic | Topic label | Proportion (%) | Top-10 words | FREX | Lift |
T10 | Deaths & Mortality | 15.1 | peopl, match, kill, game, polic, break, dozen, tonight, grow, washington | kill, peopl, break, match, grow, dozen, game | grow, kill, washington, break, peopl, dozen, tonight |
T11 | Family Impact | 8.7 | dead, fan, match, player, arm, cop, lose, wife, daughter, teenag | arm, dead, player, fan, wife, daughter, cop | arm, daughter, wife, player, teenag, dead, cop |
T15 | Condolences | 8.1 | famili, victim, happen, live, condol, lost, sad, incid, game, prayer | condol, friend, peac, love, happen, rip, live | friend, heartfelt, passion, rip, condol, deepest, extend |
T14 | Stampede Trigger | 7.6 | tear, gas, polic, fire, pitch, trigger, invad, fan, control, angri | trigger, invad, pitch, tear, respond, gas, quell | shatter, invad, respond, watchdog, trigger, quell, invas |
T9 | Riot | 5.9 | iot, injur, report, video, leav, cnn, sky, unrest, footballmatch, photo | riot, cnn, report, injur, sky, video, policesay | policesay, cnn, sky, footballmatch, riot, report, video |
T3 | FIFA President reacts | 5.8 | tragedi, sport, presid, fifa, histori, day, dark, hope, widodo, weekend | presid, dark, widodo, tragedi, joko, infantino, comprehens | comprehens, dark, expos, fix, gianni, infantino, joko |
T7 | Location | 5.3 | malang, kanjuruhan, arema, event, persebaya, clash, broke, affect, sadden, surabay | persebaya, arema, surabaya, sadden, kanjuruhan, malang, deepli | impun, surabaya, corner, persebaya, tactic, arema, hide |
T4 | Death Toll | 5.2 | death, toll, children, offici, rise, mourn, result, revis, mass, relat | toll, revis, rise, relat, reach, jump, offici | jump, ministri, toll, earlier, reach, revis, stadiumstamped |
T16 | Ban & Suspend | 4.6 | leagu, ban, support, suspend, erupt, countri, human, intern, associ, play | leagu, suspend, play, ban, fuck, bet, danger | hooligan, institut, play, activ, asap, bet, fuck |
T13 | Impose Sanctions | 4.5 | disast, world, worst, deadliest, fifa, cup, explain, pssi, host, sanction | worst, cup, world, sanction, react, disast, pressur | asham, sever, statut, cup, worst, amateur, pressur |
T6 | Police chief sack | 4.2 | olic, java, east, chief, offic, sunday, provinc, local, support, remov | chief, east, provinc, java, remov, local, offic | chief, law, provinc, remov, sack, east, capac |
T8 | Probe | 4.2 | time, investig, team, york, nyt, secur, probe, set, includ, independ | york, nyt, independ, lee, investig, muktita, sui | form, ian, lee, role, surviv, austin, brief |
T18 | Panic | 4.0 | polic, exit, trampl, caus, panic, left, gate, fire, field, suffoc | panic, escap, caus, rush, panick, soccermatch, exit | chaotic, escap, vermaut, fled, panick, disastr, drive |
T17 | Survivor experiences | 3.7 | die, chao, due, hospit, lock, contribut, horror, wit, recount, blame | lock, contribut, horror, recount, die, survivor, wit | contribut, door, lock, recount, survivor, horror, delay |
T5 | Crowd response | 3.3 | crush, crowd, respons, mount, anger, justic, punish, call, cri, left | respons, justic, crush, mount, cri, crowd, punish | cri, justic, perpetr, respons, demand, mount, move |
T2 | News headlines | 3.2 | news, bbc, stori, read, top, terribl, score, horribl, guardian, come | news, bbc, read, articl, score, headlin, quot | headlin, quot, bbc, articl, news, found, read |
T12 | Vigil | 2.6 | saturday, night, hundr, victim, silenc, minut, octob, held, vigil, hurt | silenc, minut, octob, vigil, memori, night, kick | : imag, tribut, uefa, candlelight, kick, memori, minut |
T1 | Event evolution | 2.3 | violenc, watch, author, unfold, reuter, post, inindonesia, stadiumcrush, footballstamped, fear | violenc, inindonesia, stadiumcrush, reuter, footballstamped, timelin, answer | answer, footballstamped, inindonesia, stadiumcrush, timelin, rivalri, reuter |
T19 | Sequence of events | 1.8 | match, death, occur, led, troubl, game, stand, polic, peopl, saturday | occur, match, troubl, led, death, stand, game | troubl, led, occur, death, stand, match, game |
Topic | Topic label | Proportion (%) | Top-10 words | FREX | Lift |
T10 | Deaths & Mortality | 15.1 | peopl, match, kill, game, polic, break, dozen, tonight, grow, washington | kill, peopl, break, match, grow, dozen, game | grow, kill, washington, break, peopl, dozen, tonight |
T11 | Family Impact | 8.7 | dead, fan, match, player, arm, cop, lose, wife, daughter, teenag | arm, dead, player, fan, wife, daughter, cop | arm, daughter, wife, player, teenag, dead, cop |
T15 | Condolences | 8.1 | famili, victim, happen, live, condol, lost, sad, incid, game, prayer | condol, friend, peac, love, happen, rip, live | friend, heartfelt, passion, rip, condol, deepest, extend |
T14 | Stampede Trigger | 7.6 | tear, gas, polic, fire, pitch, trigger, invad, fan, control, angri | trigger, invad, pitch, tear, respond, gas, quell | shatter, invad, respond, watchdog, trigger, quell, invas |
T9 | Riot | 5.9 | iot, injur, report, video, leav, cnn, sky, unrest, footballmatch, photo | riot, cnn, report, injur, sky, video, policesay | policesay, cnn, sky, footballmatch, riot, report, video |
T3 | FIFA President reacts | 5.8 | tragedi, sport, presid, fifa, histori, day, dark, hope, widodo, weekend | presid, dark, widodo, tragedi, joko, infantino, comprehens | comprehens, dark, expos, fix, gianni, infantino, joko |
T7 | Location | 5.3 | malang, kanjuruhan, arema, event, persebaya, clash, broke, affect, sadden, surabay | persebaya, arema, surabaya, sadden, kanjuruhan, malang, deepli | impun, surabaya, corner, persebaya, tactic, arema, hide |
T4 | Death Toll | 5.2 | death, toll, children, offici, rise, mourn, result, revis, mass, relat | toll, revis, rise, relat, reach, jump, offici | jump, ministri, toll, earlier, reach, revis, stadiumstamped |
T16 | Ban & Suspend | 4.6 | leagu, ban, support, suspend, erupt, countri, human, intern, associ, play | leagu, suspend, play, ban, fuck, bet, danger | hooligan, institut, play, activ, asap, bet, fuck |
T13 | Impose Sanctions | 4.5 | disast, world, worst, deadliest, fifa, cup, explain, pssi, host, sanction | worst, cup, world, sanction, react, disast, pressur | asham, sever, statut, cup, worst, amateur, pressur |
T6 | Police chief sack | 4.2 | olic, java, east, chief, offic, sunday, provinc, local, support, remov | chief, east, provinc, java, remov, local, offic | chief, law, provinc, remov, sack, east, capac |
T8 | Probe | 4.2 | time, investig, team, york, nyt, secur, probe, set, includ, independ | york, nyt, independ, lee, investig, muktita, sui | form, ian, lee, role, surviv, austin, brief |
T18 | Panic | 4.0 | polic, exit, trampl, caus, panic, left, gate, fire, field, suffoc | panic, escap, caus, rush, panick, soccermatch, exit | chaotic, escap, vermaut, fled, panick, disastr, drive |
T17 | Survivor experiences | 3.7 | die, chao, due, hospit, lock, contribut, horror, wit, recount, blame | lock, contribut, horror, recount, die, survivor, wit | contribut, door, lock, recount, survivor, horror, delay |
T5 | Crowd response | 3.3 | crush, crowd, respons, mount, anger, justic, punish, call, cri, left | respons, justic, crush, mount, cri, crowd, punish | cri, justic, perpetr, respons, demand, mount, move |
T2 | News headlines | 3.2 | news, bbc, stori, read, top, terribl, score, horribl, guardian, come | news, bbc, read, articl, score, headlin, quot | headlin, quot, bbc, articl, news, found, read |
T12 | Vigil | 2.6 | saturday, night, hundr, victim, silenc, minut, octob, held, vigil, hurt | silenc, minut, octob, vigil, memori, night, kick | : imag, tribut, uefa, candlelight, kick, memori, minut |
T1 | Event evolution | 2.3 | violenc, watch, author, unfold, reuter, post, inindonesia, stadiumcrush, footballstamped, fear | violenc, inindonesia, stadiumcrush, reuter, footballstamped, timelin, answer | answer, footballstamped, inindonesia, stadiumcrush, timelin, rivalri, reuter |
T19 | Sequence of events | 1.8 | match, death, occur, led, troubl, game, stand, polic, peopl, saturday | occur, match, troubl, led, death, stand, game | troubl, led, occur, death, stand, match, game |