Explainable Natural Language Processing for Corporate Sustainability Analysis

Subjectivity in sustainability reporting undermines accountability: The paper identifies that corporate sustainability assessments suffer from significant subjectivity on both sides—in the data itself (incomplete, ambiguous, or unreliable disclosures) and in how analysts evaluate them (limited resources, potential bias). This subjectivity can prevent companies that don't meet sustainability standards from facing appropriate consequences, potentially making regulations ineffective.

Corporate sustainability is inherently complex to measure: The diversity of company operations across different geographies, sizes, business types, and stakeholder relationships makes it difficult to create uniform sustainability assessments. This complexity contributes to inconsistent evaluation and makes it harder to compare sustainability performance across organizations.

Explainable AI offers a solution to transparency challenges: The paper proposes that Explainable Natural Language Processing (XN) can help address subjectivity in sustainability analysis by making the reasoning behind assessments transparent and understandable, rather than relying on opaque human judgment or "black box" AI systems.

Multiple frameworks exist but lack standardization: The paper notes that various sustainability measurement frameworks—including the UN's sustainable development goals and the global sustainability reporting framework—have been proposed, but their diversity suggests a lack of unified standards that could reduce inconsistency in how companies are evaluated.