This analysis of over 120,000 FDA Freedom of Information Act requests demonstrates how natural language processing can reveal systemic inefficiencies in public data access. By categorizing requests into distinct topics, the study highlights that response times vary drastically depending on the nature of the records, with complex regulatory filings like 510(k)s suffering from significant delays compared to simpler, previously released information. A critical implication is the discrepancy between agency expectations and actual processing speeds, particularly for medical device documentation. While the FDA warns of lengthy waits, the data shows substantial variability and persistent bottlenecks in reviewing confidential information. This suggests that current public disclosure mechanisms may be inadvertently slowed by manual redaction processes, undermining the transparency goal of open data initiatives. This article is highly relevant to open data because it proves that simple aggregation of FOIA logs is insufficient for understanding accessibility. It underscores the necessity of advanced text analysis to identify specific categories of withheld or delayed data. By exposing these granular disparities, researchers can advocate for targeted policy changes and better data release strategies, ensuring that open government data is not just available, but timely and usable for stakeholders.
Source:Published on 2023-04-06