Papers by Dominic Vincent Ligot
Social Science Research Network, 2021
This paper examines opportunities for big data in augmenting public health surveillance. We descr... more This paper examines opportunities for big data in augmenting public health surveillance. We describe the use of social listening, remote sensing and mobility indices in supplementing surveillance data. We discuss the time varying reproductive number as an alternative to the basic reproductive number for monitoring an ongoing pandemic, present the INFORM risk management model for integrating insights, and discuss modalities for decision support systems. Selected examples of big data surveillance applications are presented along with considerations for further work.
Social Science Research Network, 2022
Objectives. We developed a survival scorecard from the Titanic passenger dataset using statistica... more Objectives. We developed a survival scorecard from the Titanic passenger dataset using statistical scoring methods. We discuss the scorecard development process, assess the effectiveness of statistical scorecards, and analyze the characteristics of Titanic passengers that led to survival. Methods. From the Titanic dataset of 1,309 passengers and a binary dependent variable representing survival, we assessed nine (9) features using chi-square, Weight of Evidence (WoE), and Information Value. A logistic regression was fitted on the feature WoEs to predict survival, feature coefficients were used to determine score weights for each attribute, and an additive model on attribute scores per passenger determined survival scores. The resulting scorecards were assessed for risk ranking, and the characteristics of the passenger population were assessed for survivability and population shifts. Results. The resulting survival scorecard was able to rank survivability amongst Titanic passengers (K-S 0.558, ROC 0.892, CAP 0.765) with survival classification accuracy varying by score cutoff (AR 0.63-0.84). Sex was the strongest predictor of survivability (IV 145), followed by fare amount (IV 53), cabin class (IV 52), and passenger class (IV 51). Women passengers had four times higher survivability compared to men (72% vs. 16%). Passengers who paid $100 or more for their trip had nearly ten times higher survivability compared to free passengers (75% vs. 7.6%). Cabin passengers had higher survivability compared to non-cabin passengers with cabin B having nearly three times higher survivability compared to non-cabin passengers (75% vs. 27%). Class 1 passengers had nearly three times higher survivability compared to Class 3 passengers (62% vs. 22%). Conclusion. This paper illustrates the benefits of statistical scoring methods compared to other machine learning approaches in the analysis of event likelihood risks and performing population risk segmentation. Machine learning approaches usually focus on prediction accuracy while scorecards allow for cross-sectional analysis of population risks. Scorecard cutoffs provide avenues for decision-making on populations accounting for tradeoffs in accuracy, recall, precision, and specificity.
This paper explores the principles and practices of AI governance, underscoring its critical role... more This paper explores the principles and practices of AI governance, underscoring its critical role in ensuring the safe and responsible utilization of AI technologies. It outlines the two main types of AI systems (discriminative and generative) and the prominent model architectures driving recent advancements (transformers and latent diffusion models). The paper highlights key engineering challenges, including training data quality, prompt engineering, content misuse prevention, and acceptable use guidelines. It also delineates crucial roles in AI adoption and governance, such as builders, users, planners, and trainers. Ethical considerations surrounding generative AI are examined, emphasizing accuracy, human oversight, transparency, accountability, and fairness. The paper lists the current risks and fears of generative AI in the media. The 4E Framework is proposed as a holistic approach encompassing education, engineering practices, enforcement mechanisms, and ethical principles. Effective AI governance, through frameworks like the 4E model, is pivotal for fostering trust, transparency, and accountability in AI development, deployment, and usage while aligning with ethical standards and societal values for the benefit of all stakeholders.
This article discusses the emerging relationship of generative AI and performance, citing results... more This article discusses the emerging relationship of generative AI and performance, citing results from the MIT and Stanford study in a Philippines call center, a BCG experiment with GPT-4, and University of Minnesota's randomized controlled trial amongst their law school students. We then discuss the recent surveys of the Philippines' ranking in interest and usage of generative AI tools. We detail the need for skills across four profiles: Builders, Users, Leaders, and Trainers of AI. Finally we discuss the need for ethics in promoting adoption of generative AI tools.
Social Impact Research Lab examines opportunities for analytics and artificial intelligence in cr... more Social Impact Research Lab examines opportunities for analytics and artificial intelligence in creating social impact. It presents frameworks and methodologies that utilize data for social good. Existing applications and publications on analytics and artificial intelligence that were crafted within the realm of social good are presented along with the discussion of its current and potential impact on society.
Objectives. The paper aims to perceive if a country's cultural characteristics following Hofstede... more Objectives. The paper aims to perceive if a country's cultural characteristics following Hofstede's Cultural Dimensions have an impact on its educational performance, in general. The study intends to: (i) determine the relationship between educational performance and cultural dimensions; (ii) evaluate how each cultural dimension affects educational performance; and (iii) present possible prediction of PISA scores by evaluating cultural dimensions following Hofstede's Cultural Dimensions Theory.
Objectives. This paper aims to perceive if there is a trade-off between the time spent on the int... more Objectives. This paper aims to perceive if there is a trade-off between the time spent on the internet and social media and the educational performance of students, across countries. Specifically, this study intends to determine the relationship between internet usage and educational performance; evaluate the impact of social media usage to educational performance; and predict PISA scores by evaluating data on internet and social media usage. Methods. The experiment utilizes PISA scores as dependent variable while time spent on internet and social media are the independent variables. Correlation analysis was performed to assess the relationship among the target variables. Linear regression was executed to generate a model that predicts PISA scores using data on time spent on the internet and social media. Results. The results show a strong inverse relationship between the dependent variable (PISA scores) and the independent variables (time spent on internet and/or social media). Model fitting results using 2018 data show that the variability of the model with time spent on social media as the predictor is significantly higher (62% R-squared) than with time spent on the internet as the predictor (24% R-squared). Highest variance explanation (74% R-squared) is constituted on the model with time spent on both internet and social media as predictors. Discussion and Recommendations for Future Research. The results of the experiment strongly suggest that the time spent on internet and social media generally has negative impact on PISA scores. Notably, the experiment also showed that the independent variables are potentially good predictors of PISA scores. The resulting coefficients of the third scenario imply that time spent on the internet marginally increases PISA score, but time spent on social media decreases it. Interestingly, the predictive power of time spent on social media was stronger in 2018 than 2015 perhaps due to the fact that in moving towards a digital world, the use of social media is more prevalent as years pass. This study may need a follow-through when the next round of PISA results become available. By then, the impact of pandemic could be embedded into the analysis.
Air quality has been observed to improve during the COVID-19 pandemic as countries executed commu... more Air quality has been observed to improve during the COVID-19 pandemic as countries executed community lockdowns that restricted mobility during the pandemic. This paper discusses the Mobility over Air Quality Index (MAQI) which aims to gauge relative changes in levels of air pollution generated by changes in population mobility. MAQI combines mobility indices produced by Google with NO2 readings obtained from Earth Observations as a metric that can be visualized over time. We discuss the development of MAQI from the recent EO Dashboard Hackathon sponsored by NASA, ESA, and JAXA, recent developments in the project, and directions for further study.
A new era for Europe. Volume II, Emerging challenges, 2023
The COVID-19 pandemic, which hit the EU in January 2020, brought unprecedented challenges and mea... more The COVID-19 pandemic, which hit the EU in January 2020, brought unprecedented challenges and measures to contain the contagion. Difficult and unpopular decisions were taken to limit European citizens’ freedom of movement, assembly, and simply meeting each other. Economic activities in many sectors were suspended. Lockdowns were imposed. Yet, all these measures could only slow down the spread of the virus, and thousands still died in the space of months. Vaccines were developed in late 2020 and started to be administered, not without shortcomings, in early 2021. However, public opinion was strained by then with successive waves of the virus, recurrent restrictions and obligations to wear masks. Protest movements arose in many EU Member States by mid-2021, particularly against vaccination and mask-wearing obligations. In this context, it is reasonable to expect that the trust in governments and public institutions would decline. This paper explores changes in trust in governments during the COVID-19 pandemic, using real-time open-source data. The research presented aimed at identifying changes in trust in governments throughout the crisis, determining if these can be linked to the pandemic development and containment and if they have the potential to impact electoral results. Two case studies, Spain and Germany, were selected for their diverging experiences of the pandemic. Public discussions on social media were searched for relevant conversations related to the trust in political leaders as proxies for governments, as posts on the social media are usually personalised, naming leaders rather than referring to institutions. These posts were then analysed for sentiment, particularly trust. Finally, topics were modelled to establish whether the social media posts were related to the pandemic and its management. The results that the research brought are limited, given the focus on just two case studies, but encouraging both on their findings and on the method, which allows for inexpensive, real-time analysis of public opinion, particularly at the time of crisis when surveys may not be practical.
The Ambit, 2023
This paper is a petition that highlights the urgent need for effective governance of Artificial I... more This paper is a petition that highlights the urgent need for effective governance of Artificial Intelligence (AI) in the Philippines. It addresses the concerns and risks associated with the rapid advancement of AI capabilities and emphasizes the importance of inclusive and comprehensive policies to mitigate potential harms and foster responsible AI development. The petition draws attention to significant events, such as the call to pause giant AI experiments, the US Federal Trade Commission complaint against OpenAI, and the concerns raised by deep learning pioneer Geoffrey Hinton, which underscore the global apprehension surrounding AI. Furthermore, the paper emphasizes the specific concerns of AI workers in the Majority World, with a focus on the Philippines, and the role of the country in formulating timely regulations. Recommendations are proposed to enhance AI governance, including the development of a national AI strategy, the establishment of an AI ethics committee, adherence to international standards like the UNESCO Recommendation on Ethics of AI, promotion of AI literacy and upskilling programs, protection of vulnerable groups, and the consideration of environmental sustainability. By implementing these recommendations, the Philippines can contribute to the ethical governance of AI and safeguard the interests of its citizens in the digital era. The paper underscores the need for collaboration with key stakeholders and civil society organizations to ensure transparency, accountability, and public participation in AI policymaking processes.
JEL: C18, O33, O38, K23, L86, Q55
Conflict Studies: Prevention, Management & Resolution eJournal, 2022
We describe the Bangsamoro Data Challenge, a platform for collaboration, ideation, and solution d... more We describe the Bangsamoro Data Challenge, a platform for collaboration, ideation, and solution development to solve social issues in the Bangsamoro Autonomous Region of Muslim Mindanao (BARMM) the Philippines. The challenge invited data enthusiasts, students, and working professionals to create solutions by utilizing the publicly-available datasets such as government, satellite, climate, health, agriculture, and local data. We discuss the history, relevance, process and impact of this platform, as well as the current status and potential of the prototypes generated by the exercise. The success of the challenge has paved the path towards data-driven development of the BARMM.
Decision-Making Models & Tools eJournal, 2022
Objectives. We developed a survival scorecard from the Titanic passenger dataset using statistica... more Objectives. We developed a survival scorecard from the Titanic passenger dataset using statistical scoring methods. We discuss the scorecard development process, assess the effectiveness of statistical scorecards, and analyze the characteristics of Titanic passengers that led to survival. Methods. From the Titanic dataset of 1,309 passengers and a binary dependent variable representing survival, we assessed nine (9) features using chi-square, Weight of Evidence (WoE), and Information Value. A logistic regression was fitted on the feature WoEs to predict survival, feature coefficients were used to determine score weights for each attribute, and an additive model on attribute scores per passenger determined survival scores. The resulting scorecards were assessed for risk ranking, and the characteristics of the passenger population were assessed for survivability and population shifts. Results. The resulting survival scorecard was able to rank survivability amongst Titanic passengers (K-S 0.558, ROC 0.892, CAP 0.765) with survival classification accuracy varying by score cutoff (AR 0.63-0.84). Sex was the strongest predictor of survivability (IV 145), followed by fare amount (IV 53), cabin class (IV 52), and passenger class (IV 51). Women passengers had four times higher survivability compared to men (72% vs. 16%). Passengers who paid $100 or more for their trip had nearly ten times higher survivability compared to free passengers (75% vs. 7.6%). Cabin passengers had higher survivability compared to non-cabin passengers with cabin B having nearly three times higher survivability compared to non-cabin passengers (75% vs. 27%). Class 1 passengers had nearly three times higher survivability compared to Class 3 passengers (62% vs. 22%). Conclusion. This paper illustrates the benefits of statistical scoring methods compared to other machine learning approaches in the analysis of event likelihood risks and performing population risk segmentation. Machine learning approaches usually focus on prediction accuracy while scorecards allow for cross-sectional analysis of population risks. Scorecard cutoffs provide avenues for decision-making on populations accounting for tradeoffs in accuracy, recall, precision, and specificity.
Philippine Business for Education, 2022
This labor market intelligence report provides a holistic overview of the supply, demand, and mis... more This labor market intelligence report provides a holistic overview of the supply, demand, and mismatch of skills in the Analytics labor sector of the Philippines. With the aim of informing skills trends and supporting growth of the labor market amidst the Fourth Industrial Revolution coupled with implications brought by the global pandemic, this report also presents an initial attempt in extrapolating the Philippine analytics workforce, with backcasted and forecasted projections from 2010 to 2028. Through a mixed-methods research, the study examined various quantitative, qualitative, and big data sources to understand the interplay of supply and demand for skills, and to provide corresponding key insights and recommendations intended to guide the Analytics Association of the Philippines, as the established Skills Sector Council, create an inclusive skills development roadmap. The report highlights the need to standardize the definitions of Analytics roles, leveraging the framework proposed by the Analytics Association of the Philippines. We discuss the need for more specialized Analytics courses, the production of more instructors, Analytics as a distinct sector from IT-BPM, and the prospect of professional licensing and certification for the sector. We also highlight existing trends that promote the development of the Analytics labor sector such as women participation, work from home arrangements, online learning, the emergence of Analytics communities, and the impending importance of Data and AI Ethics.
UNDP Philippines, 2021
Thus, UNDP Philippines, in close collaboration with the National Economic and Development Authori... more Thus, UNDP Philippines, in close collaboration with the National Economic and Development Authority (NEDA), has commissioned this research titled, “Trends in COVID-19 Vaccine Acceptance in the Philippines and Their Implications on Health Communication”, to deepen our understanding of the factors behind vaccine acceptance in Philippines.
In this research, we applied innovative methodologies to generate insights for community mobilization and social behavior change communication (or SBCC) interventions, which could be an effective strategy in addressing vaccine acceptance. The report generated significant insights related to the level of vaccine acceptance, factors that determine the change in behavior and identified strategic communication messaging cues.
Academia Letters, Aug 12, 2021
This paper presents Project AEDES, a big data early warning, and surveillance system for dengue. ... more This paper presents Project AEDES, a big data early warning, and surveillance system for dengue. The project utilizes Google Search Trends to detect public interest and panics related to dengue. Using Google Search Trends, precipitation, and temperature readings from climate data, the system nowcasts probable dengue cases and dengue-related deaths. The system utilizes FAPAR, NDVI, and NDWI readings from remote sensing to detect likely mosquito hotspots to prioritize interventions. We discuss the origin and development of the project and recent developments. We also discuss the current state of development and directions for further work.
SSRN Applied Computing eJournal, 2021
This paper examines opportunities for big data in augmenting public health surveillance. We descr... more This paper examines opportunities for big data in augmenting public health surveillance. We describe the use of social listening, remote sensing and mobility indices in supplementing surveillance data. We discuss the time varying reproductive number as an alternative to the basic reproductive number for monitoring an ongoing pandemic, present the INFORM risk management model for integrating insights, and discuss modalities for decision support systems. Selected examples of big data surveillance applications are presented along with considerations for further work.
SSRN INFORMATION SYSTEMS: BEHAVIORAL & SOCIAL METHODS eJOURNAL, 2021
Objectives. Infodemics of false information on social media is a growing societal problem, aggrav... more Objectives. Infodemics of false information on social media is a growing societal problem, aggravated by the occurrence of the COVID-19 pandemic. The development of infodemics has characteristic resemblances to epidemics of infectious diseases. This paper presents several methodologies which aim to measure the extent and development of infodemics through the lens of epidemiology.
Methods. Time varying R was used as a measure for the infectiousness of the infodemic, topic modeling was used to create topic clouds and topic similarity heat maps, while network analysis was used to create directed and undirected graphs to identify super-spreader and multiple carrier communities on social media.
Results. Forty-two (42) latent topics were discovered. Reproductive trends for a specific topic were observed to have significantly higher peaks (Rt 4-5) than general misinformation (Rt 1-3). From a sample of social media misinformation posts, a total of 385 groups and 804 connections were found within the network, with the largest group having 1,643 shares and 1,063,579 interactions over a 12 month period.
Conclusions. These approaches enable the measurement of the infectiousness of an infodemic, comparative analysis of infodemic topics, and identification of likely super-spreaders and multiple carriers on social media. The results of these analyses can form the basis for taking action to stem an ongoing spread of misinformation on social media and mitigate against future infodemics. The methods are not confined to health misinformation and may be applied to other infodemics, such as conspiracy theories, political disinformation, and climate change denial.
SSRN POLITICAL BEHAVIOR: VOTING & PUBLIC OPINION eJOURNAL, 2021
Objectives. Public trust is a key determinant of public health policies and risk-reduction strate... more Objectives. Public trust is a key determinant of public health policies and risk-reduction strategies during a pandemic. This study aims to assess the public trust towards government responses in different countries during the COVID-19 pandemic, and to determine the relationship between socio-demographic factors and public trust.
Methods. We conducted an online survey using convenience sampling between 25 March 2020 to 31 March 2020 to measure public trust in government response during the COVID-19 pandemic in different countries using a questionnaire adapted from a previous study in 2009 during the H1N1 outbreak. We also investigated the relationship between socio-demographic characteristics and the Trust Score using multivariate analyses, and compared the Trust Scores between countries to distinguish countries with different levels of public trust towards the government.
Findings. Responses were collected from 87 countries. Only 7 out of 87 countries surveyed (with at least 30 respondents) were included in further analyses. Among the 7 countries selected for comparison, respondents from India and Malaysia have the highest levels of public trust towards government responses, while public trust is the lowest in the United States. The data collected from 2 countries with at least 350 responses (India and Malaysia) also showed the socio-demographic factors did not predict Trust Scores at a statistically significant level.
Conclusion. Respondents in India and Malaysia have high levels of public trust, and the level of public trust is low in the United States. Socio-demographic factors failed to predict public trust at a statistically significant level.
Uploads
Papers by Dominic Vincent Ligot
JEL: C18, O33, O38, K23, L86, Q55
In this research, we applied innovative methodologies to generate insights for community mobilization and social behavior change communication (or SBCC) interventions, which could be an effective strategy in addressing vaccine acceptance. The report generated significant insights related to the level of vaccine acceptance, factors that determine the change in behavior and identified strategic communication messaging cues.
Methods. Time varying R was used as a measure for the infectiousness of the infodemic, topic modeling was used to create topic clouds and topic similarity heat maps, while network analysis was used to create directed and undirected graphs to identify super-spreader and multiple carrier communities on social media.
Results. Forty-two (42) latent topics were discovered. Reproductive trends for a specific topic were observed to have significantly higher peaks (Rt 4-5) than general misinformation (Rt 1-3). From a sample of social media misinformation posts, a total of 385 groups and 804 connections were found within the network, with the largest group having 1,643 shares and 1,063,579 interactions over a 12 month period.
Conclusions. These approaches enable the measurement of the infectiousness of an infodemic, comparative analysis of infodemic topics, and identification of likely super-spreaders and multiple carriers on social media. The results of these analyses can form the basis for taking action to stem an ongoing spread of misinformation on social media and mitigate against future infodemics. The methods are not confined to health misinformation and may be applied to other infodemics, such as conspiracy theories, political disinformation, and climate change denial.
Methods. We conducted an online survey using convenience sampling between 25 March 2020 to 31 March 2020 to measure public trust in government response during the COVID-19 pandemic in different countries using a questionnaire adapted from a previous study in 2009 during the H1N1 outbreak. We also investigated the relationship between socio-demographic characteristics and the Trust Score using multivariate analyses, and compared the Trust Scores between countries to distinguish countries with different levels of public trust towards the government.
Findings. Responses were collected from 87 countries. Only 7 out of 87 countries surveyed (with at least 30 respondents) were included in further analyses. Among the 7 countries selected for comparison, respondents from India and Malaysia have the highest levels of public trust towards government responses, while public trust is the lowest in the United States. The data collected from 2 countries with at least 350 responses (India and Malaysia) also showed the socio-demographic factors did not predict Trust Scores at a statistically significant level.
Conclusion. Respondents in India and Malaysia have high levels of public trust, and the level of public trust is low in the United States. Socio-demographic factors failed to predict public trust at a statistically significant level.
JEL: C18, O33, O38, K23, L86, Q55
In this research, we applied innovative methodologies to generate insights for community mobilization and social behavior change communication (or SBCC) interventions, which could be an effective strategy in addressing vaccine acceptance. The report generated significant insights related to the level of vaccine acceptance, factors that determine the change in behavior and identified strategic communication messaging cues.
Methods. Time varying R was used as a measure for the infectiousness of the infodemic, topic modeling was used to create topic clouds and topic similarity heat maps, while network analysis was used to create directed and undirected graphs to identify super-spreader and multiple carrier communities on social media.
Results. Forty-two (42) latent topics were discovered. Reproductive trends for a specific topic were observed to have significantly higher peaks (Rt 4-5) than general misinformation (Rt 1-3). From a sample of social media misinformation posts, a total of 385 groups and 804 connections were found within the network, with the largest group having 1,643 shares and 1,063,579 interactions over a 12 month period.
Conclusions. These approaches enable the measurement of the infectiousness of an infodemic, comparative analysis of infodemic topics, and identification of likely super-spreaders and multiple carriers on social media. The results of these analyses can form the basis for taking action to stem an ongoing spread of misinformation on social media and mitigate against future infodemics. The methods are not confined to health misinformation and may be applied to other infodemics, such as conspiracy theories, political disinformation, and climate change denial.
Methods. We conducted an online survey using convenience sampling between 25 March 2020 to 31 March 2020 to measure public trust in government response during the COVID-19 pandemic in different countries using a questionnaire adapted from a previous study in 2009 during the H1N1 outbreak. We also investigated the relationship between socio-demographic characteristics and the Trust Score using multivariate analyses, and compared the Trust Scores between countries to distinguish countries with different levels of public trust towards the government.
Findings. Responses were collected from 87 countries. Only 7 out of 87 countries surveyed (with at least 30 respondents) were included in further analyses. Among the 7 countries selected for comparison, respondents from India and Malaysia have the highest levels of public trust towards government responses, while public trust is the lowest in the United States. The data collected from 2 countries with at least 350 responses (India and Malaysia) also showed the socio-demographic factors did not predict Trust Scores at a statistically significant level.
Conclusion. Respondents in India and Malaysia have high levels of public trust, and the level of public trust is low in the United States. Socio-demographic factors failed to predict public trust at a statistically significant level.