AI against Modern Slavery: Digital Insights into Modern Slavery Reporting - Challenges and Opportunities Nyasha Weinberg,1 Adriana Bora,2 Francisca Sassetti,3 Katharine Bryant,4 Edgar Rootalu,5 Kar- yna Bikziantieieva,6 Laureen van Breen,7 Patricia Carrier,8 Yolanda Lannquist,9 Nicolas Miailhe10 The Future Society,1,2,5,6,9,10 Walk Free,3,4 WikiRate,7 Business and Human Rights Resource Centre8 fsassetti@walkfree.org,3 adriana.bora@thefuturesociety.org,2 kbryant@walkfree.org,4 laureen@wikirate.org,7 carrier@business-human- rights.org8 Abstract Keywords: Artificial Intelligence, AI for Good, Modern Slavery, Business Due Diligence, Human Rights, Supply From seafood from Thailand and electronics from Malaysia Chain Ethics. and China, to textiles from India and wood from Brazil, mod- ern slavery exists in all corners of the planet. It is a multi- billion-dollar transnational criminal business that affects us all through trade and consumer choices. In 2016, an estimated Introduction 25 million people were forced to work through threats, vio- lence, coercion, deception, or debt bondage. Of these, 16 mil- From seafood from Thailand and electronics from Malaysia lion were forced to work in the private sector. Given the and China, to textiles from India, wood from Brazil, and ap- widespread nature of the problem, governments, corpora- parel manufacturing in the United Kingdom,1 modern slav- tions, and the general public are increasingly expecting com- ery exists in all corners of the planet. Modern slavery is a panies to accurately disclose the actions they are taking to multi-billion-dollar transnational criminal business that af- tackle modern slavery. Yet, five years on, there are chal- lenges with understanding companies’ compliance under the fects us all through trade and consumer choices. In 2016, an 2015 UK Modern Slavery Act. It is unclear which companies estimated 25 million people were forced to work through are failing to report under the MSA, while the quality of these threats, violence, coercion, deception, or debt bondage. Of statements often remains poor. Project AIMS (Artificial In- these, 16 million were forced to work in the private sector telligence against Modern Slavery) harnesses the power of (ILO and Walk Free 2017). It is estimated that approxi- artificial intelligence (AI) for tackling modern slavery by an- alyzing modern slavery statements to assess compliance with mately US$354 billion worth of products at-risk of being the UK and Australian Modern Slavery Acts, in order to produced by forced labor are imported by G20 countries an- prompt business action and policy responses. This paper ex- nually (Walk Free 2018). Given the widespread nature of amines the challenges and opportunities for better machine the problem, governments, corporations, and the general readability of modern slavery statements identified in the in- public are increasingly expecting companies to accurately itial stages of this project. Machine readability is important to extract data from modern slavery statements to enable disclose the actions they are taking to tackle modern slav- analysis using AI techniques. Although extensive technolog- ery.i A valuable source of information is corporate reporting ical solutions can be used to extract data from PDFs and resulting from supply chain transparency requirements in HTMLs, establishing transparency and accessibility require- domestic legislation.2 ments would reduce the resources required to assess modern slavery reporting and ultimately understand what companies are doing to address modern slavery in their direct operations and supply chains - unlocking this critical ‘AI for Social Good’ use case. AAAI Fall 2020 Symposium on AI for Social Good. giant Boohoo, where workers received significantly less than minimum Copyright © 2020 for this paper by its authors. Use permitted under Crea- wage and worked without protective equipment (Duncan 2020; Matety tive Commons License Attribution 4.0 International (CC BY 4.0). 2020). 2 See UK Modern Slavery Act 2015, Australian Modern Slavery Act 2018, 1 A recent undercover investigation brought to light the slavery-like exploi- California Supply Chain Transparency Act 2010, French Duty of Vigilance tative conditions in a factory in Leicester producing clothes for fashion Law 2017. The Future Society,3ii in partnership with Walk Free,4iii This publication addresses the challenges and opportuni- launched Project AIMS (Artificial Intelligence against Mod- ties identified during the first phase, namely the process of ern Slavery) in May 2020. Project AIMS seeks to, firstly, accessing, gathering and structuring the data from the com- understand how we can harness the power of artificial intel- pany statements. Data collection and structuring is a key ligence (AI) to increase the efficiency of assessing compli- cornerstone in building any successful AI project and thus ance with the UK and Australian Modern Slavery Acts. Sec- this publication puts forward a set of lessons learned and ondly, the project will allow us to understand how we can recommendations on good practices to facilitate the applica- harness the power of AI for policymaking by providing ac- tion of AI for Social Good. This paper adopts the perspec- tionable insights for governments, businesses, and civil so- tive that although extensive technological solutions can be ciety organizations. The overarching project will attempt to used to extract data from modern slavery statements in PDF identify and share best practices in modern slavery report- and HTML formats, establishing transparency and accessi- ing, and identify specific sectors where reporting is falling bility requirements would reduce the resources required to short. It will make recommendations for companies on how do so. It focuses on changes that would enable resource-con- to improve compliance with the UK and Australian Modern strained technical experts to extract data in a more efficient Slavery Acts and for governments considering developing manner than what is technically feasible today. similar legislation on how to maximize its impact. Project AIMS builds upon the work of Walk Free, WikiRate,5iv and Business & Human Rights Resource Cen- Background tre (BHRRC)6v to assess the statements produced under the Following California’s 2010 Transparency in Supply Chains UK Modern Slavery Act. It draws from the BHRRC Modern Act, the UK developed the first national legal framework for Slavery Registry to develop an AI algorithm to ‘read’ and transparency in supply chains: the 2015 Modern Slavery assess the statements produced by companies under supply Act. It includes a provision that requires companies supply- chain transparency legislation.vi This algorithm will use 18 ing goods or services in the UK with an annual turnover of metrics designed by Walk Free in line with the UK Home £36 million or more to publish an annual modern slavery Office guidance (UK Government 2017) to assess state- statement indicating the steps they are taking to identify and ments, and will be integrated with the WikiRate platform to address modern slavery risks. enable ongoing human verification of the automated data Yet, five years on, there are several challenges in under- collection. standing business compliance with the UK Modern Slavery There are four phases to Project AIMS. The first phase of Act. It is difficult to establish which companies are failing Project AIMS is focused on accessing, gathering and struc- to report, while the variable quality of the statements re- turing the data from existing company statements, building leased makes it difficult to understand the actions companies the largest publicly available text corpus of modern slavery are taking to address modern slavery. With an estimated 12 statements.7 In phase two of the project, we will design an 000-17 000 UK-based companies having to publish state- automated labeling function through weak supervision tools ments per annum, few studies have attempted to assess these to increase the amount of available labeled data. Once suffi- reports due to the laborious nature of manually analyzing cient data are correctly labeled, the third phase of Project each statement (Walk Free et al. 2019). For example, even AIMS begins: using supervised machine learning methods the most comprehensive study to date, conducted by Walk to create a document classifier, which can assess modern Free and WikiRate (2018), sampled just over 900 reports slavery statements against the 18 metrics. Lastly, in the and took almost two years to complete As companies con- fourth phase, the results will be published, and the tool will tinue to report under the UK legislation and start to report be made publicly available through an open-source API. under the 2018 Australian Modern Slavery Act, failure to 3 practices. By bringing this information together in one place, and making The Future Society is an independent 501(c)(3) nonprofit think-and-do it accessible, comparable and free for all, the organization provides society tank working on advancing the responsible adoption of AI and other emerg- with the tools and evidence it needs to spur companies to respond to the ing technologies for the benefit of humanity. world's social and environmental challenges. To date, WikiRate.org is the 4 Tackling one of the world’s largest and most complex human rights issues largest open source registry of ESG data in the world, with currently almost requires serious strategic thinking. Walk Free approaches this challenge by 900,000 data points for over 55 000 companies. 6 integrating world class research with direct engagement with some of the The BHRRC is an international, non-profit organization that works to ad- world’s most influential government, business, and religious leaders. We vance human rights in business and eradicate abuse. Its website tracks the invest our time and resources in a collaborative manner to drive behavior activities of more than 10 000 companies around the world. 7 and legislative change to impact the lives of the estimated 40 million people This corpus combines the small amount of “labeled statements” (the mod- living in modern slavery today. ern slavery statements manually benchmarked against the 18 metrics by 5 volunteers from WikiRate and Walk Free) with the large amount of “unla- WikiRate is a nonprofit that hosts an open data platform which allows beled statements” (the statements from the Modern Slavery Registry that anyone to systematically gather, analyze and report publicly available in- have not yet been benchmarked). formation on corporate Environmental, Social and Governance (ESG) address these obstacles to efficiently and consistently assess AIMS seeks to demonstrate that, beyond optimizing busi- modern slavery statements will undermine the potential of ness performance, the use of AI-based solutions can be lev- this legislation to improve transparency and accountability eraged to strengthen the rule of law, specifically supply in business operations and supply chains. chain transparency legislations that address modern slavery To date, access to modern slavery statements has been risk, remediation, and prevention. through company websites8 or the compilation efforts by the BHRRC’s Modern Slavery Registry,vii TISC,viii and WikiRate,ix who have collected, collated, and analyzed these Recommendations data. Much of this information has been collated manually, with teams of researchers searching for, and systematically Based on the creation of the dataset under the first phase of reviewing, available statements. Given this is a costly exer- Project AIMS, we set out the following recommendations cise that requires a lot of man-hours, a more centralized and for policymakers and companies to improve access to mod- automated approach is desirable. Promising steps in this re- ern slavery reporting using technology. gard are the development of the UK Home Office registry, and the recent launch of Australia's registry, which will cen- For Policymakers tralize the housing of these statements.x Technological inno- vations will also reduce the time taken to extract relevant Recommendation 1: Governments with modern slavery re- information from these statements. This enables insights porting requirements should publish an up-to-date, compre- into company disclosure of actions to remove modern slav- hensive list of all companies and their subsidiaries subject ery from their operations and supply chains, and also facili- to reporting. tates the automation of elements of the assessment of these Recommendation 2: Governments should keep a single reg- statements. istry where companies must submit their statements. These This is a technically challenging task. However, the chal- statements must have consistent formatting to ensure easy lenges in dealing with large complex structured and unstruc- retrieval. tured data sets are not new, and neither is the quest to har- 2a Ensure all statements are required to disclose the report- ness AI technologies to tackle them (Pferd 2010).9 Big data ing period, are timestamped, and include relevant has been widely adopted as a solution to tackle the mam- metadata,10 such as the address of company headquarters, moth task of exploring and extracting meaningful insights that assist interoperability with other data sources. from large structured and unstructured datasets (Adnan and 2b In addition to housing on the company homepage, require Akbar 2019; Rai 2017; Yang et al. 2019). There are also ev- companies to submit their statements to the registry. ident gaps in data governance and the need for a more holis- 2c House historical statements in this same registry. tic view to guide both practitioners and researchers in this Recommendation 3: Governments should legislate that field (Abraham, Schneider, and vom Brocke 2019). Promi- companies should publish statements in machine readable nent areas of this application include the medical and formats11 to improve comparability and support transpar- ency. healthcare sectors, with several studies showing how the use of AI to structure data sets and extract information can con- tribute to the prevention of infectious diseases and identifi- For Companies cation of key areas of interventions, but not without its own Recommendation 1: Companies subject to modern slavery challenges (Cohen et al. 2017; McCue and McCoy 2017). reporting requirements should endeavor to assist govern- This project aims to use AI to support the achievement of ments with keeping an up-to-date, comprehensive list of all the Sustainable Development Goals (SDG), including SDG companies and their subsidiaries subject to reporting. 8 aimed at: Recommendation 2: Companies should place their modern slavery statements on their homepage, with a URL that in- Promot[ing] sustained, inclusive and sustainable eco- cludes the reporting year. nomic growth, full and productive employment and de- cent work for allxi 2a Ensure that all statements disclose the reporting period, are timestamped, and include relevant metadata that assists 8 10 Examples of best practice under the UK Modern Slavery Act (Business & Based on our research to-date, a good metadata for this purpose would Human Rights Resource Centre 2018); Examples of publications under the be the address of company headquarters. 11 Duty of Vigilance Law: (Carrefour 2018). A machine-readable format is a type of structured format that can be read 9 and processed by a computer. Examples suitable for modern slavery state- It is important to note that many important data management and analyt- ments include Extensible Markup Language (XML). A machine-readable ics tasks cannot be full done by automated processes, therefore crowdsourc- format does not include PDF, although different PDF formats facilitate ing is used to harness human cognitive abilities to process some computer readability. tasks, such as sentiment analysis and image recognition. This area of work has been extensively studied in recent years as Li et al. (2017) suggest. interoperability with other data sources, such as the address across two separate data sets of modern slavery statements of company headquarters. from the WikiRate platform and the Modern Slavery Regis- 2b In addition to housing on their homepage, submit their try. While these databases have collected statements pub- statements to the registry, with consistent formatting. lished by companies, they are inevitably incomplete due to 2c Provide records of historical statements on their website the inherent difficulties of collecting all statements in scope. and in the registry. The goal of this comparison was to conduct a gap analysis Recommendation 3: Companies should publish in a ma- and assist with the identification of additional statements. chine-readable format, with infographics and images com- This analysis has revealed the difficulty of analyzing com- prehensively explained in text that fully summarizes and ref- panies with complex structures, often with multiple subsid- erences all information contained within. iaries, inconsistent industry classifications, and companies that span multiple industries, which creates challenges for generating a comprehensive streamlined dataset of compa- Challenges nies. Accessing high-quality, structured, machine readable data b) Scraping reports from company websites from companies’ Modern Slavery Act statements is a signif- Based on the analysis by Project AIMS, from the approxi- icant challenge (Rodriguez 2018). This is particularly true mately 17 000 unique statement URLs stored in the Modern when assessing a large number of these statements to iden- Slavery Registry, just 12 005 could be accessed. In approx- tify sector-specific characteristics, or to illustrate change imately 4 913 cases, errors blocked the scraping process, of over time. However, detailed reports are not mandatory, nor which 328 errors were related to HTML stored formats. The are these statements standardized or saved in consistent for- remaining 4 585 errors affected those stored in PDF format. mats. The content included in statements is left to the dis- More precisely, of the approximately 10 700 URLs contain- cretion of companies, resulting in vast differences in sub- ing the statements in PDF format, only 6 212 statements stance and quality. This presents several problems for the could be accessed. use and development of AI to facilitate the extraction and These errors were caused by a number of issues that make analysis of relevant information at scale. website scraping complicated, such as shifting webpage structures, redirects and CAPTCHAS,12 unclear navigation, Access Challenges and unstructured HTML.13 These issues include: • Statement missing from homepage. Not all companies fol- Access is a significant issue facing anyone who wants to ex- low government requirements to publish modern slavery tract data. Data are accessible for AI if it can be identified, statements in a prominent place on their homepage (Home extracted, processed, and parsed easily by a computer. Office 2019). • Shifting webpage structures. Website redesign means that Identification of Relevant Reports sections can become more complicated to access. To extract data from relevant reports requires the identifica- • Connection issues caused by URL structures. Complicated tion of companies that are subject to mandatory reporting URL structures, including multiple query strings and requirements. In the UK, this process currently requires vis- hashes create significant connection issues. iting company websites and drawing from existing datasets • Connection issues caused by server connection errors. To (such as WikiRate or the Modern Slavery Registry), as there scrape the statements, the computer sends a request for is no centralized government registry yet. This raises a num- processing to the web server that hosts the statement, and ber of issues, including: the server then sends a response to the computer running the code. If the server is not connected, it is not possible a) Finding companies that are subject to a reporting to scrape data. duty • Unclear links. Some links direct to a page which hosts a To date, there is no publicly available list of companies number of links to modern slavery statements instead of which are in scope of the UK Modern Slavery Act. This the most recent statement itself. This leads to the text be- makes it incredibly difficult, if not impossible, to identify ing extracted from that website instead of the text from which companies are in scope of the Act, and pinpoint which the actual reports.xii While it is helpful for companies to have a webpage that links to all of the previous modern should have reported, but have not yet done so. The AIMS slavery statements in one place, it is essential for project compares metadata variables (e.g. ‘name’ or ‘URL’) 12 A CAPTCHA is a type of test to determine whether or not a particular 13 Unstructured HTML is where the HTML has not been tagged in a con- user is human. sistent pattern that allows for analysis. Sometimes, unstructured HTML is simply a consequence of bad programming (Kansal 2019). automation purposes to have the most up-to-date state- ment in an easily traceable location. • Blocked scraping. Some websites block instances when text is scraped multiple times. While this may be useful in some contexts, when applied to a page that houses modern slavery statements it hampers transparency. Format of Reporting The format of modern slavery reporting can also add hurdles for the extraction of data. Lack of Digital Formats A new EU regulation requires all financial statements to be published in a digital format (Laermann 2018). The UK Modern Slavery Act, on the other hand, does not mandate companies to publish modern slavery statements in a single Figure 1. Example risk assessment heatmap. Source: authors. electronic format. This means that the statements are incon- sistent and different approaches need to be taken to extract data from each format. Figure 1 demonstrates some of these challenges. If the in- formation contained within the heatmap was captured Extraction of Data from PDFs within a paragraph text, the tool could easily extract the in- Data published in PDF format, which is the form that mod- formation “Bananas and prawns are the products most at ern slavery statements often take, is not easily machine read- risk." It is possible to use computer vision to read the text, able (Pollock 2016), which makes it more difficult to iden- but without additional code to read the colors as risk indica- tify, read, extract and analyze information automatically. tors we would not be able to rank the information contained Specific issues include: within the figure. • Scanned PDFs. Scanned PDFs are often not machine read- able as they are captured as a solid image. Optical Char- • Sub-formats of PDF. Each specific format of PDF requires acter Recognition (OCR) can help conversion into ma- a separate OCR solution for extracting the data. chine-encoded text that can then be analyzed, but this is more arduous than using a PDF saved directly from a computer. • Formatting. The use of formatting, including borders, mul- tiple columns, inconsistent column widths, pop-out boxes, headers, and footers, adds to the complexity of data extraction. • Data embedded in images and graphics. A further chal- lenge is extracting data that are embedded in images and graphics., which present challenges to the structuring and automatic processing of data. Without developing spe- cialized methods for extracting data from complex tables, figures, and graphs, these can scramble the information contained within. Based on the analysis so far, out of the 5 903 extracted statements in HTML format, 96 state- ments have data embedded in meaningful images, while out of 6,092 statements in PDF format, 237 contained meaningful images.14 14 A meaningful image is any kind of infographics containing information format, supply chain embedded in the image, description of the company that is important for the benchmarking of a metric (e.g. report in image etc. This does not include images containing signatures). Figure 2. Example of diagram. Images and diagrams can make text There is also an opportunity to extend financial reporting extraction difficult. Source: authors. requirements to modern slavery reporting and ESG data to assist efforts to source and efficiently integrate data into cross-asset investment decisions and implementation. Com- Figure 2 also provides important information on a com- panies’ annual financial reports are made machine-readable pany’s modern slavery strategy, but the use of a diagram under new European Securities and Markets Authority rules. creates additional difficulties in the process of reading, ex- Doing the same with modern slavery statements and ESG tracting and structuring data. These diagrams are, however, data would improve comparability, support transparency essential for a number of stakeholder groups to help them and contribute to increased investor protection (Rust 2017). understand company modern slavery strategies, which is At a minimum, structured reporting, even if not in XHTML why we do not recommend removing them, but rather sup- format, would align with the EU’s 2013 Transparency Di- plementing them with a text-based description. rective Recital 26 which states that: Structure of Reports a harmonised electronic reporting format would be very beneficial for issuers, investors and competent au- • Section Titles. Without clearly demarcated section head- thorities, since it would make reporting easier and fa- ings that mirror the government’s sections for reporting, cilitate accessibility, analysis and comparability of an- it can be difficult to find relevant information for specific nual financial reports (European Securities and metrics. Markets and Authority n.d.). • Alignment with reporting standards. Use of conventional terminology enables easier extraction, and further analy- sis would be enhanced if this aligned with globally spe- Given the challenges and opportunities for machine read- cific reporting standards or frameworks (e.g. SASB, EU ability of modern slavery reporting explored in this paper, frameworks). we believe that establishing transparency and accessibility • Tagging. Labels by companies to assist machine readabil- requirements would reduce the resources required to assess ity would be very helpful; this is particularly important in modern slavery reporting, increase understanding of the ac- areas where we see inconsistent typologies used by com- tions companies are taking to address modern slavery, and panies to describe similar phenomena. This could follow ultimately hold companies accountable for the exploitation suggested or mandated criteria from governments.15 that occurs in their direct operations and in their supply chains. Opportunities References Using Structured, Machine Readable Formats Abraham, Rene, Johannes Schneider, and Jan vom Brocke. 2019. across Corporate Reports “Data Governance: A Conceptual Framework, Structured Review, and Research Agenda.” International Journal of Information Man- Machine-readable formats would make information con- agement 49:424–38. doi: 10.1016/j.ijinfomgt.2019.07.008. tained in modern slavery statements more easily accessible, Adnan, Kiran, and Rehan Akbar. 2019. “An Analytical Study of which would facilitate data retrieval and allow for better Information Extraction from Unstructured and Multidimensional comparisons between companies within and across sectors Big Data.” Journal of Big Data 6(1):91. doi: 10.1186/s40537-019- and countries, and show change over time. It would also al- 0254-8. low for adjustments that enable access for people with disa- Australian Government Department of Home Affairs. 2018. Com- bilities. monwealth Modern Slavery Act 2018 - Guidance for Reporting En- The methodology developed to extract relevant infor- tities. mation through Project AIMS could also be applied to other Business & Human Rights Resource Centre. 2018. “Unilever Dis- reporting frameworks to develop a comprehensive picture of closes Entire Palm Oil Supply Chain; Explains Decision as Vital corporate disclosure and activity. In particular, there is an in Addressing Deforestation and Human Rights Abuses.” Business & Human Rights Resource Centre, February 20. opportunity to apply this extraction technology to Environ- mental, Social and Governance (ESG) reporting frame- Carrefour. 2018. “2.6.2 Le Plan de Vigilance Du Groupe Carre- four.” works. There are currently more than 230 sustainability re- porting frameworks, which ultimately impairs rather than Cohen, Trevor, Kirk Roberts, Anupama E. Gururaj, Xiaoling Chen, Saeid Pournejati, George Alter, William R. Hersh, Dina Demner- aids the extraction, comparability, and analysis of the wealth Fushman, Lucila Ohno-Machado, and Hua Xu. 2017. “A Publicly of information contained within these reports (XBRL 2018). Available Benchmark for Biomedical Dataset Retrieval: The 15 For example, guidance by the Australian Government Department of Home Affairs (2018) identifies seven mandatory criteria for reporting which can be clearly tagged in headings across the statements. Reference Standard for the 2016 BioCADDIE Dataset Retrieval Pferd, Jeffrey W. 2010. “The Challenges of Integrating Structured Challenge.” Database 2017. doi: 10.1093/database/bax061. and Unstructured Data.” Technical Paper. Landmark. Duncan, Conrad. 2020. “Boohoo ‘Facing Modern Slavery Investi- Pollock, Rufus. 2016. “Tools for Extracting Data and Text from gation’ after Report Finds Leicester Workers Paid as Little as £3.50 PDFs - A Review.” Open Knowledge Labs. Retrieved October 19, an Hour.” The Independent, July 5. 2020 (http://okfnlabs.org). European Securities and Markets and Authority. n.d. “European Rai, Rajeshwari. 2017. “Intricacies of Unstructured Data.” ICST Single Electronic Format.” Transactions on Scalable Information Systems 4:153151. doi: Home Office. 2019. “Publish an Annual Modern Slavery State- 10.4108/eai.25-9-2017.153151. ment.” GOV.UK. Retrieved October 19, 2020 Rodriguez, Arturo. 2018. “It Takes Two to Make Data Coverage (https://www.gov.uk/guidance/publish-an-annual-modern-slav- Go Right.” SASB. Retrieved October 19, 2020 ery-statement). (https://www.sasb.org/blog/it-takes-two-to-make-data-coverage- ILO, and Walk Free. 2017. Global Estimates of Modern Slavery: go-right/). Forced Labour and Forced Marriage. Geneva, Switzerland: Inter- Rust, Susanna. 2017. “Company Financial Reports to Be Made national Labour Office (ILO). Machine-Readable.” IPE. Retrieved October 19, 2020 Kansal, Satwik. 2019. “Advanced Python Web Scraping: Best (https://www.ipe.com/company-financial-reports-to-be-made-ma- Practices & Workarounds.” Codementor Blog. Retrieved October chine-readable/10022424.article). 19, 2020 (https://www.codementor.io/blog/python-web-scraping- UK Government. 2017. Transparency in Supply Chains. A Practi- 63l2v9sf2q). cal Guide. Laermann, Michael. 2018. “Machine-Readable Disclosure and Walk Free. 2018. Global Slavery Index. ESEF 2020 – A New Era of Corporate Digital Reporting?” Sus- Walk Free, WikiRate, Business & Human Rights Resource Centre, tainable Brands. Retrieved October 19, 2020 (https://sustainable- and Australian National University. 2019. Beyond Compliance in brands.com/read/marketing-and-comms/machine-readable-disclo- the Hotel Sector: A Review of UK Modern Slavery Act Statements. sure-and-esef-2020-a-new-era-of-corporate-digital-reporting). WikiRate, and Walk Free Foundation. 2018. Beyond Compliance: Li, Guoliang, Yudian Zheng, Ju Fan, Jiannan Wang, and Reynold The Modern Slavery Act Project. Cheng. 2017. “Crowdsourced Data Management: Overview and Challenges.” Pp. 1711–1716 in Proceedings of the 2017 ACM In- XBRL. 2018. “CRD Launch Project to Standardise ESG.” XBRL. ternational Conference on Management of Data, SIGMOD ’17. Retrieved October 19, 2020 (https://www.xbrl.org/news/crd- New York, NY, USA: Association for Computing Machinery. launch-project-to-standardise-esg/). Matety, Vidhathri. 2020. “Boohoo’s Sweatshop Suppliers: ‘They Yang, Longzhi, Jie Li, Noe Elisa, Tom Prickett, and Fei Chao. Only Exploit Us. They Make Huge Profits and Pay Us Peanuts.’” 2019. “Towards Big Data Governance in Cybersecurity.” Data- The Sunday Times, July 5. Enabled Discovery and Applications 3(1):10. doi: 10.1007/s41688-019-0034-9. McCue, Molly E., and Annette M. McCoy. 2017. “The Scope of Big Data in One Medicine: Unprecedented Opportunities and Challenges.” Frontiers in Veterinary Science 4. doi: 10.3389/fvets.2017.00194. i vii Business and Human Rights Resource Centre, modernslaveryregistry.org/. Business and Human Rights Resource Centre, modernslaveryregis- Accessed October 1, 2020. try.org. Accessed October 1, 2020. ii viii The Future Society, thefuturesociety.org/. Accessed October 1, 2020. TISC Report, tiscreport.org/. Accessed October 1, 2020. iii ix Walk Free, walkfree.org/. Accessed October 1, 2020. WikiRate, wikirate.org/. Accessed October 1, 2020. iv x WikiRate, wikirate.org/. Accessed October 1, 2020. Australian Modern Slavery Registry, modernslaveryregister.gov.au/. Ac- v Business and Human Rights Resource Centre, business-human- cessed October 1, 2020. xi rights.org/en/. Accessed October 1, 2020. United Nations, sdgs.un.org/goals/goal8. Accessed October 1, 2020. vi xii Business and Human Rights Resource Centre, modernslaveryregis- Victoria and Albert Museum, vam.ac.uk/info/reports-strategic-plans- try.org. Accessed October 1, 2020. and-policies. Accessed October 1, 2020.