November 2026 Summaries

2 posts from Unstructured

Filter
Month: Year:
Post Summaries Back to Blog
No summary generated yet.
Nov 19, 2026 1,094 words in the original blog post.
Unstructured’s Transform parse API now offers an optional image-description enrichment that uses a vision model to generate prose explanations for detected figures, making chart and map findings retrievable alongside surrounding document text. Using examples from a National Hurricane Center report on Hurricane Helene, the feature identifies relationships that ordinary text extraction misses, including a histogram’s right-skewed distribution, mean and median speeds, Helene’s approximately 28 mph landfall speed, and a storm-track map’s intensity progression, landfall region, and pressure reading. Printed chart values are intended to be returned exactly where possible, while values inferred from chart geometry are approximate and better suited to retrieval than precise computation. Users can also provide a schema in the same parse request to receive structured fields such as chart titles, axis labels, statistics, and annotated points in addition to descriptions. Because the enrichment analyzes every detected figure, it may increase processing time, particularly in figure-heavy documents, and is recommended where visual content contains key findings.
Nov 12, 2026 1,066 words in the original blog post.