Contents
  1. Purpose and development
    1. What document understanding preserves
    2. From printed characters to document models
  2. Inputs and representation
    1. Inspect the available content
    2. Retain document relationships
  3. Text and page structure
    1. Prepare images and preserve coordinates
      1. Returning a highlight to the source
    2. Recognize character sequences
    3. Find regions and their roles
    4. Recover reading order and continuity
  4. Traceability
    1. Locate supporting source material
  5. Document architectures
    1. Combine text, layout and images
      1. Learning from words and positions
    2. Generate structure from page images
  6. Tables and records
    1. Recover table meaning
    2. Extract fields with context
      1. What the application added
    3. Check records against their sources
  7. Quality and recovery
    1. Measure the required fidelity
    2. Choose what to accept
    3. Recover incomplete and disputed results
      1. Review and correction
  8. Downstream use
    1. Deliver context with the records
  9. Reference
    1. Frequently asked questions
      1. Does a searchable PDF still need OCR?
      2. Can OCR-free extraction choose the wrong number?
      3. Why can a table value need several source locations?
      4. Can a partially processed document supply a usable record?
  10. Check understanding
  11. Open questions
  12. Selected talks
  13. References
  14. Talk library
← All topics

Document understanding & OCR

Document understanding is the process of turning a document into data while preserving the relationships that give its contents meaning and the source locations where a reader can check them. Optical character recognition (OCR) transcribes writing from images. In Adit Abraham’s World’s Fair 2026 example from Reducto, a transcript could reproduce Rev, US and EU exactly. To tell that the US and EU columns both report revenue, software must also parse the table, identifying its cells and the header they share. Jerry Liu of LlamaIndex explains why even a digitally authored Portable Document Format (PDF) file can require this work. Instructions that position letters and draw lines can reproduce the table on screen while leaving the relationships among its cells implicit.

Purpose and development

What document understanding preserves

The center panel of Abraham’s Formatting Strategy slide at 11:31 shows how HTML records the relationship. In the markup, Q spans two rows, so it still occupies the first column in the second header row. Rev spans the next two columns. That places the second row’s US and EU headers under Rev. The slide also shows a simple quarter-and-revenue table in Markdown at left and a prose description for retrieval at right.

These operations can run in sequence. The input determines how to recover the words. The consumer’s question determines what else to recover. A schema names the fields and types the consumer expects.
OperationConcrete inputOutputVerification
Native text extractionEncoded text on a PDF invoiceStrings and available page positionsCompare copied characters with the visible print.
OCRA scan of a printed priceA transcript of the digits and decimal markInspect the marks at the recognized location.
Layout and table reconstructionPositioned words, alignment and ruling linesCells, rows and header spansTrace each regional column to its parent heading.
Field extractionAn invoice and a line-item schemaOne record per item with its description, quantity and priceCheck that all three values belong to that item.

The same table can serve two consumers. A question about how revenue changed over time might share few words with a grid of numbers. Abraham describes indexing a natural-language description to help find the table, then supplying its structural HTML when a model needs to read the values. Liu makes a separate decision about effort. Prepare a collection economically for bulk lookup, then return to the source for detailed extraction and citations when a particular field matters. A system can use both designs.

For a searchable archive, recovering a distinctive phrase may be enough. For data entry, an invoice total and the balance after a deposit answer different questions even when both amounts are read correctly. Source fidelity means agreement with what the document says, including the labels and qualifications that affect its meaning. Faithful conversion can preserve an author’s factual mistake just as accurately as a correct statement.

The invoice total is 220 USD, but a 50 USD deposit leaves 170 USD due. Extracting the balance requires identifying the field’s role, not just recognizing a number.

From printed characters to document models

An early commercial reader had a narrowly defined job. Farrington’s machine, sold in 1956, read printed numbers on credit-sales invoice cards and punched the information into those same cards for existing equipment. Controlling the input was one way to make recognition practical. OCR-A and OCR-B were typefaces developed for optical reading in the 1960s. ISO 1073 standardized their character shapes in 1976. Exception handling also belonged to the machinery of data entry. A 1969 IBM guide describes the 1287 retaining successfully read characters and displaying unreadable ones for keyboard correction. Smithsonian collection record · OCR font history · IBM guide

A blind reader choosing a book had no control over the printer’s typeface. Kurzweil Computer Products combined a flatbed scanner, recognition across typefaces, known as omnifont recognition, and speech synthesis in a reading machine announced in January 1976. Imperfect print posed another problem. During Tesseract’s development at HP from 1984 to 1994, its recognizer used clearly read words on a page to adapt a classifier before revisiting uncertain words. Page layout came from a separate component. HP released the engine as open source in late 2005. Kurzweil’s account · Ray Smith’s account of Tesseract

Digital document interchange developed along a parallel path. At Adobe, John Warnock’s 1991 Camelot proposal addressed the difficulty of reproducing an application document on a computer with different software and fonts. PDF could package font shapes and explicit page drawing operations. That gave software access to encoded page content and made recovering characters from a scan unnecessary for some files. Reproducing a two-column page still left open how a program should read it aloud. Warnock’s oral history

Learning from complete transcripts reduced the need to mark individual character boundaries. Graves and colleagues introduced their sequence-alignment method in 2006 and tested it on speech. In 2013, Breuel and colleagues applied it to printed lines, including touching characters. Interpreting fields posed a further problem. Yiheng Xu and colleagues at Microsoft trained LayoutLM with recognized words and their page positions in a 2019 preprint published at KDD in 2020. Geewook Kim and colleagues at NAVER developed Donut to generate structured content from page images, first reported in 2021 and published at ECCV in 2022. These systems start with different evidence. Choosing among them begins with inspecting the file. Printed-line recognition · LayoutLM · Donut

Inputs and representation

Inspect the available content

A born-digital document originates in software. Its file may contain text, drawings, form fields or tags that describe logical structure. A scan supplies pixels. A searchable scan can look identical to the original image while supplying an invisible OCR transcript when someone selects text. Even one page can be mixed. If a typed heading sits above a scanned form, selecting the heading tells you little about the entries below it. Compare the text available in the target region with the rendered page before choosing an extraction path.

A glyph is the shape a font draws. A PDF can draw the right glyph and return the wrong Unicode character when copied because drawing and character extraction use different mappings. Inspect a disagreement before replacing the text. OCRmyPDF’s release notes describe a defect that left an old OCR layer inside an embedded page object and added a second reading. Replacing text can also invalidate tags pointing to the old content. When usable native text covers only part of a page, retain it and recognize the uncovered writing. PDF mapping diagnosis · OCRmyPDF processing modes

Choosing an extraction path within a page

Example

One searchable page can require different operations in different regions.

Inspect region against renderCheck both coverage and character fidelity.
Keep encoded text
Recognize the writing
Recognize gaps and combineRetain the location of each reading.
Inspect mappings and layers
Compare the rendered target with the available text. Keep faithful text, recognize image-only writing, combine complementary readings, or investigate disagreement.
Read the diagram as text
  • Inspect region against render. Check both coverage and character fidelity.
  • Keep encoded text.
  • Recognize the writing.
  • Recognize gaps and combine. Retain the location of each reading.
  • Inspect mappings and layers.
  • Inspect region against render → Keep encoded text: Faithful coverage.
  • Inspect region against render → Recognize the writing: Image-only writing.
  • Inspect region against render → Recognize gaps and combine: Partial faithful coverage.
  • Inspect region against render → Inspect mappings and layers: Conflicting readings.

Native structure can extend beyond the visible body. WordprocessingML encodes paragraphs and runs of text with shared properties, while comments, headers and footnotes can occupy separate package parts. Inventory the pages and parts in scope against an identified source version. Record unsupported or unreadable material so that a missing part cannot pass downstream as an empty one. Preserving the source snapshot establishes the start of its provenance, the record of origins and transformations. WordprocessingML structure · Origins, times and permitted use

Retain document relationships

An intermediate document representation gives software identifiable objects to work with before it exports strings or records. A page can contain regions, a region can contain lines, and a line can contain words. Containment answers where an item belongs in this hierarchy. Consider a page with two pictures and a caption. All three share the same page parent. Following a separate link from the caption to one picture tells the program which picture it describes. A model-generated description can be attached to the same picture while remaining distinguishable from the author’s printed caption.

A bounding box is a rectangle locating an item in a specified coordinate frame. DoclingDocument can attach boxes to items, order the children of its body tree in reading sequence, and keep repeating page furniture such as footers in a separate tree. An item reference addresses this conversion. Retaining the source version and spatial location makes it possible to inspect the item even if another conversion gives it a different identifier.

Which picture does the caption describe?

Example

A relationship between siblings carries meaning that their shared parent cannot express.

Page
First picture
Second picture
Printed caption
The page contains both pictures and the caption. Following the caption’s describes link identifies just the second picture.
Read the diagram as text
  • Page.
  • First picture.
  • Second picture.
  • Printed caption.
  • Page → First picture: contains.
  • Page → Second picture: contains.
  • Page → Printed caption: contains.
  • Printed caption → Second picture: describes.

Plain text exports one sequence. Markdown can preserve headings and simple tables. CSV supplies rectangular records. Keep the original, observed text, normalized values, extraction method and completion state available behind those exports. Established OCR formats illustrate several ways to retain more. hOCR embeds spatial text in HTML. ALTO XML can record order and reference processing steps for individual elements. PAGE XML represents polygonal regions and can group their identifiers in ordered or unordered sets. hOCR specification · Library of Congress ALTO documentation · PAGE XML example

Text and page structure

Prepare images and preserve coordinates

Rasterization renders a page into pixels at a chosen resolution. Inspect the smallest mark that can change the reading, such as a decimal point. Rotate a sideways page to its intended orientation. Deskew tilted lines so the recognizer can group their letters. Thresholding classifies pixels as foreground or background and can separate ink from uneven paper, but it can also remove faint strokes. Dewarping straightens print on a curved surface by moving different parts of the page different distances. Retain both the source render and the image actually sent to recognition. Tesseract’s input-quality guidance

Returning a highlight to the source

In the two pixel frames, follow region R. The crop starts at (40, 20) on the source. Subtract that origin from R’s upper-left corner at (50, 30) to get (10, 10) within the crop. Doubling the crop places the corner at (20, 20) in the recognition image. The opposite corner moves from (70, 40) to (60, 40). To highlight the result on the source, divide both recognized coordinates by two and add the crop origin. If padding was added after resizing, subtract the padding first.

One region before and after cropping and resizing

Example

The crop origin moves while the region doubles in pixel dimensions.

Source render

A 50 by 30 pixel crop encloses R.

027.55582.5110020406080Horizontal position (pixels)Vertical position (pixels)Region RCrop boundaryCrop originR
  • 1. Crop boundary
  • 2. Region R
Read coordinates and regions as data

X: 0–110 pixels; Y: 0–80 pixels, increasing down. Equal scale on both axes.

Crop boundary (polyline)

(40, 20); (90, 20); (90, 50); (40, 50); (40, 20)

Region R (polygon)

(50, 30); (70, 30); (70, 40); (50, 40)

Crop origin: (40, 16)

R: (60, 36)

Recognition image

The crop now starts at (0, 0) and measures 100 by 60 pixels.

027.55582.5110020406080Horizontal position (pixels)Vertical position (pixels)Region RCrop boundaryR
  • 1. Crop boundary
  • 2. Region R
Read coordinates and regions as data

X: 0–110 pixels; Y: 0–80 pixels, increasing down. Equal scale on both axes.

Crop boundary (polyline)

(0, 0); (100, 0); (100, 60); (0, 60); (0, 0)

Region R (polygon)

(20, 20); (60, 20); (60, 40); (20, 40)

R: (40, 31)

The dashed boundary encloses a crop beginning at (40, 20). Subtracting its origin and doubling both coordinates maps R from (50, 30)–(70, 40) to (20, 20)–(60, 40). Each panel measures pixels from its own upper-left origin.

Use the actual transform for each page. Rotation requires mapping all four corners of a word box. Nonlinear dewarping requires a spatial map. Coordinates expressed as fractions of a frame need that frame’s dimensions. A PDF highlight may require conversion from image pixels to page points, accounting for the visible crop and page rotation. A location can map back accurately even when resizing has changed the pixel samples or cropping has discarded context. PyMuPDF page geometry · Computer Vision

To test whether smoothing helps recognition, compare the transcripts with and without smoothing on the same source images. Character error rate counts the character substitutions, deletions and insertions needed to align a transcript with a reference, divided by the reference length. On 25 images with Tesseract 4.1.0 in its default configuration, the DewarpNet authors reported a rate of 0.249 without added blur and 0.3136 with it. Smoothing made the text harder for that recognizer to read on those images. DewarpNet evaluation notes

Recognize character sequences

Detection locates writing, grouping supplies a word or line image, and recognition assigns it a transcript. A character classifier that receives two touching letters as one cutout must work around an error made before classification. A sequence recognizer can learn from the transcript of a whole line without an annotated cut between each pair of letters. It still needs an appropriate region and enough detail to distinguish the marks.

Figures 1 and 2 in Shi, Bai and Yao’s paper show a cropped word becoming a sequence of columns of visual features. Their convolutional recurrent neural network (CRNN) uses neighboring columns on both sides to interpret each position. It can assign a character or a special blank label at each position. Connectionist Temporal Classification (CTC) trains the recognizer by summing probabilities over paths of labels compatible with the target transcript. A wide character can occupy several successive positions without requiring a person to specify its width. CRNN figures · Original CTC paper

Two runs produce two digits

Example

Removing a blank too early would turn two digit runs into one.

1 / 3 · Position labels

Two adjacent positions predict the first digit.

New this step1 · 1 · blank · 1One possible label path across the image.
In the path 1, 1, blank, 1, collapse adjacent repeated labels to get 1, blank, 1. Removing blank then produces 11.
Read the diagram as text
  • 1 · 1 · blank · 1. One possible label path across the image.
  • 1 · blank · 1. One label per consecutive run.
  • Transcript 11.
  • 1 · 1 · blank · 1 → 1 · blank · 1: collapse adjacent repetitions.
  • 1 · blank · 1 → Transcript 11: remove blank.
  1. Position labels. Two adjacent positions predict the first digit. Active: 1 · 1 · blank · 1. New: 1 · 1 · blank · 1.
  2. Consecutive runs. The blank keeps the last digit in its own run. Active: 1 · 1 · blank · 1, 1 · blank · 1. New: 1 · blank · 1.
  3. Transcript. The two digit runs now contribute one character each. Active: 1 · 1 · blank · 1, 1 · blank · 1, Transcript 11. New: Transcript 11.

To read a possible path, collapse consecutive identical labels and then remove blanks. In the figure, the blank between two runs of 1 allows the transcript to contain two identical digits. A lexicon, a set of permitted strings, can help choose among uncertain readings. A dictionary of ordinary words can also exclude the correct product code. When a result is implausible, inspect the crop for blur, compression damage or unusual print, and check whether the recognizer supports the script and handwriting on the page. Even an exact reading of a code leaves its role to be determined.

Find regions and their roles

Physical segmentation groups marks into regions. Role assignment identifies what each region does, such as introducing a section, presenting a table or qualifying a figure. A mistaken boundary can send the wrong material to the next stage. Group a paragraph with a neighboring table and its sentences become candidates for cell contents. Drop a small footnote region and even a correctly located reference mark has nowhere to point.

Top-down partitioning starts with a page. Ha, Haralick and Phillips’s 1995 recursive X-Y cut projects the rectangles enclosing connected marks onto each axis. A sufficiently large empty interval permits a horizontal or vertical cut, and the procedure repeats inside each resulting rectangle. On the partitioned pages in the paper, look for whitespace extending all the way across the region being divided. A gutter that stops at a spanning heading cannot separate the columns in a single cut through that heading. Recursive X-Y cut

Bottom-up grouping starts with neighboring marks. Mao and Kanungo’s implementation study contrasts Docstrum’s ability to group nonrectangular arrangements with its difficulties when characters are sparse or irregularly spaced. Learned detectors can use visual features to predict region boundaries and roles, but even their reference labels require judgment. In IBM’s DocLayNet Figure 4, inspect the choice between enclosing subfigures together and identifying them separately. An annotation convention settles what the detector should learn. Keeping nontext and unknown regions addressable leaves a later stage somewhere to attach a caption. Segmentation study · DocLayNet Figure 4

Recover reading order and continuity

Reading order specifies the precedence that preserves a document’s meaning. In W3C’s two-column PDF example, follow a line to the gutter. The sentence continues on the next line of the same column. Sorting all words by height and then horizontal position would instead cross into the other column. The words stay correct while the sentence changes. Tagged PDFs can supply a reading order directly. Following the tags across the same boundary tests whether conversion preserved the author’s sequence. W3C reading-order example

Constructed document example. Reading down each column preserves the two procedures. Sorting all lines by their vertical position would interleave the before-printing and after-printing steps.

A magazine page may contain two independent articles. Each opening precedes its own continuation, but the reader can choose either article first. A partial order records just the precedence relations that must hold. A plain-text export selects one traversal. Keep an explicit link from a note to the passage it qualifies so the association survives whichever traversal the exporter chooses. Reading-order evaluation

Precedence within independent articles

Example

A note can qualify a passage without determining which article comes first.

Article A opening
Article A continuation
Article B opening
Article B continuation
Note
Each opening precedes its own continuation. Neither article has priority over the other. The note points to the passage whose interpretation it changes.
Read the diagram as text
  • Article A opening.
  • Article A continuation.
  • Article B opening.
  • Article B continuation.
  • Note.
  • Article A opening → Article A continuation: precedes.
  • Article B opening → Article B continuation: precedes.
  • Note → Article A continuation: qualifies.

Within a line, text direction poses a different problem. Arabic or Hebrew can include left-to-right digits and Latin text. Unicode distinguishes a string’s logical character order from the order of glyphs on display. Across pages, a logical paragraph can continue past a footer and a repeated header. A readable export can omit that furniture from the paragraph and join a word split by a line-ending hyphen. Preserve links to both original fragments when making the join. Unicode bidirectional algorithm · Tagged PDF guide

Traceability

Locate supporting source material

A source location addresses material in a particular document version. An evidence span identifies the text or regions that support a reading. Consider a duration exported from a table. One region contains the digits, while a header above several columns supplies the unit. For verification, open both regions on the archived page and check that the header governs this cell. A link to today’s file at the same URL or a search for the first occurrence of the digits can resolve to different evidence.

The address must identify what its numbers count.
AddressInformation needed to resolve it
Physical page indexSource version and numbering convention. A display label can repeat, use Roman numerals or differ from the number printed on the page.
Box or polygonPage or image frame, units, origin and any crop or rotation. An oriented quadrilateral can follow tilted text more precisely than an axis-aligned box.
Text spanExact text representation, counting unit and boundary convention. W3C’s position selector includes the start and excludes the end.
OCR word ID or model token positionThe conversion or tokenized input in which the identifier resolves, plus a mapping back to source text. Tokenization

Two locations support one duration

Example

Inspecting the number resolves only one branch of the field’s evidence.

Duration field
Numeric text span
Unit text span
Data cell region
Governing header region
Page in archived sourceIdentified by source version and physical index.
The magnitude resolves to a cell and the unit to a governing header. Both regions must resolve on the identified source page.
Read the diagram as text
  • Duration field.
  • Numeric text span.
  • Unit text span.
  • Data cell region.
  • Governing header region.
  • Page in archived source. Identified by source version and physical index.
  • Duration field → Numeric text span: magnitude read from.
  • Duration field → Unit text span: unit read from.
  • Numeric text span → Data cell region: located in.
  • Unit text span → Governing header region: located in.
  • Data cell region → Page in archived source: region of.
  • Governing header region → Page in archived source: region of.

Expanding fi into fi makes two output characters correspond to one source character. Joining a word across lines can give one output word two source regions. ICU’s edit map records corresponding ranges even when the interior of a replacement has no exact character-level alignment. For reordered blocks, retain the original segments in the new traversal order. Google Document AI’s document resource illustrates a useful separation between text anchors into a string and page anchors into polygons. A model can propose either kind of anchor. Inspecting the addressed material establishes what it supports. ICU edit maps · Document AI resource

Document architectures

Combine text, layout and images

A converter can obtain a word’s spelling from one source and its cell assignment from another. In Docling’s original native-PDF pipeline, one branch extracts encoded words with coordinates. Another renders the page, identifies table regions and predicts cells, spans and header roles using TableFormer. At the point where the branches meet, post-processing assigns words to predicted cells by position. Document assembly then incorporates the populated table alongside other elements in reading order. If a price lands one column over, inspect the predicted boundary and the word assignment before rerunning recognition. Docling technical report

In Cedric Clyburn’s recorded notebook, Clyburn of Red Hat converts Docling’s own research paper, inspects its pages and renders extracted tables separately. To check a table, choose a value on the source page, identify its row and column heading, then perform the same lookup in the extracted table. Finding the digits somewhere in the Markdown export would test a weaker claim.

Assigning native words to predicted cells

A correct spelling can enter the wrong cell where the text and image branches meet.

PDF
Positioned words
Rendered page
Page regions and roles
Table structure recognitionTableFormer predicts cells, spans and roles.
Assign text to cells
Document assembly
In Docling’s native-PDF route, positioned words come from encoded text. Layout and table models recover cell geometry from page imagery. Matching the words to cells produces the populated table for assembly.
Read the diagram as text
  • PDF.
  • Positioned words.
  • Rendered page.
  • Page regions and roles.
  • Table structure recognition. TableFormer predicts cells, spans and roles.
  • Assign text to cells.
  • Document assembly.
  • PDF → Positioned words: extract text and coordinates.
  • PDF → Rendered page: render page pixels.
  • Rendered page → Page regions and roles: predict regions.
  • Page regions and roles → Table structure recognition: table boundaries.
  • Rendered page → Table structure recognition: pixels inside boundaries.
  • Table structure recognition → Assign text to cells: predicted cells.
  • Positioned words → Assign text to cells: spellings and positions.
  • Assign text to cells → Document assembly: populated table.
  • Page regions and roles → Document assembly: regions and roles.

Learning from words and positions

Layout-aware models learn how position changes the interpretation of words. During pretraining, when a model learns representations for later tasks, the original LayoutLM used recognized words and scaled page coordinates. Its architecture added separately obtained image features downstream. Its form experiment labeled spans. Correctly labeling a name and an address still leaves their association to be checked. LayoutLMv2 brought imagery into pretraining while continuing to consume OCR words. Its tasks included aligning text lines to image regions and detecting mismatched text and images. In a comparison of base models trained for one pass on the same corpus, adding visual information and the additional objectives improved document-question-answering validation scores. LayoutLM Figure 2 · LayoutLMv2 experiments · Pretraining and Midtraining

Generate structure from page images

Donut takes a page image and generates a sequence encoding content and structure. Figure 3 pairs a receipt with emitted item records. This route removes the external OCR dependency. Reading and field assignment still occur within the model. In one published failure, Donut returned a different visible number for the publication number. Its attention visualization suggested where the decoder looked, without certifying field coordinates. Larger images helped with tiny text at additional computational cost. Donut figures and experiments

A vision-language model processes images and language together. IBM’s Granite-Docling accepts whole-page and regional instructions and can generate DocTags, structural markers convertible into a DoclingDocument. Working on a region can give a small field more pixels, but a tightly cropped value may lose its governing header. Google’s Gemini 3 PDF interface supplies native PDF text alongside visual page information. Compare routes on the actual document types by inspecting what each sees, which intermediate readings can be checked, and whether its returned locations resolve to the necessary context. Granite-Docling model card · Gemini document inputs · Multimodal Models and Applications

Tables and records

Recover table meaning

Table detection finds the area. Structure recognition identifies rows, columns, empty cells and cells spanning several positions. Rules drawn on the page can suggest boundaries, while a borderless table can express the same grid through alignment. Transcription supplies cell contents. Functional analysis identifies which cells act as headers for others. Microsoft’s Table Transformer exposes the separation directly. Models detect tables and recover their structure from images, but the exporter needs separately supplied text from native extraction or OCR. PubTables-1M also addressed a problem in training references, where annotations sometimes split one spanning header into several cells. Table Transformer · PubTables-1M

In the Journal Article Tag Suite (JATS) example, columns inherit their unit from a spanning header. To interpret 151/597, follow the cell’s footnote reference. JATS table and footer · W3C multi-level headers

Constructed document example. The value 5 inherits North from its row, Standard from its column, days from a spanning header, and a customs-delay qualification from the footnote. Exporting the number alone loses those relationships.

Inherited units and a referenced qualification

Header spans apply across columns. Note references apply to individual cells.

Header giving days as the unit
First subordinate column
Second subordinate column
151/597A cell elsewhere in the table.
Data unavailable for one trialA note in the table footer.
In the JATS table, one header supplies days to two columns. In a separate part of the table, the fraction 151/597 references a note about unavailable data from one trial.
Read the diagram as text
  • Header giving days as the unit.
  • First subordinate column.
  • Second subordinate column.
  • 151/597. A cell elsewhere in the table.
  • Data unavailable for one trial. A note in the table footer.
  • Header giving days as the unit → First subordinate column: supplies unit.
  • Header giving days as the unit → Second subordinate column: supplies unit.
  • 151/597 → Data unavailable for one trial: references note.

A repeated header after a page break can introduce more rows of the same table. Microsoft’s template guidance labels rows across expected page breaks as one table and calls for training examples covering those variations. Check unfamiliar continuations against the pages. Preserve empty cells that hold a column position, cell regions, header spans and note links. A simple grid can go directly to Markdown. A complex table can use HTML spans as in Reducto’s example. For CSV, expand inherited headings into column names and carry qualifications separately so a reader of the export can recover them. Template model guidance

Extract fields with context

Field extraction assigns evidence to a named role within a scope. An Address label can appear once under supplier and again under customer. Position, section membership and the label together identify the requested address. Amazon Textract distinguishes CHILD links, such as a line containing words, from VALUE links connecting a key with its value. A selection mark in a rating grid takes its meaning from both the row question and the answer column. Detecting an unchecked box supports a negative selection. Failing to detect the box leaves its state unknown. Textract relationships · Selection elements

For repeated records, extract related fields together or retain enough evidence to join them. Ben Kus describes Box dividing large extraction requests while keeping contract parties with their addresses. Independently sorted lists could contain every correct name and every correct address, yet pairing them by index could assign a party someone else’s address. A complete line item likewise requires the description, quantity and price to refer to the same product.

What the application added

Keep the literal observation accessible when changing a value for use in an application. The evidence needed to reproduce the change depends on the operation. Google’s receipt examples · Cleaning is semantic
OperationExampleRetain
NormalizationGoogle’s French receipt date 11/12/2022 becomes 2022-12-11.Literal span, location and date convention. Decimal separators and units also need an explicit interpretation. Leave alternatives open when the convention is unknown.
CorrectionA reviewer changes a misread character in an identifier.Original reading, visible mark, amendment and reviewer.
CalculationSubtract a deposit from the invoice total to calculate the balance.Both amounts, their units and source regions, and the calculation used.
EnrichmentAn external lookup supplies an address missing from a receipt.Printed mention and separate origin of the added address.

Check records against their sources

A schema can require a numeric amount for each line item. Peppol rule BR-CO-13 checks a relationship among those amounts. To obtain the invoice total excluding value-added tax, add the net amounts for all lines, subtract document-level allowances and add document-level charges. Comparing the result with the total read independently from the page can expose a missing line. If the extractor instead fills the total field from its own shortened list, the two figures will agree. Even an independently read total cannot expose a swap of two line net amounts because the sum stays the same. Check each amount against its source row to verify which item it belongs to. Peppol BR-CO-13 · What a schema promises · Structure is not domain correctness

For a correctly typed price attached to the wrong item, the decisive check must inspect the pairing.
CheckEvidence inspectedWhat can pass unnoticed
StructureRecord fields and typesAn amount copied from another row
ReconciliationAmounts supplied to the arithmetic ruleExchanging two amounts
AssociationSource row, label and value togetherAnother row omitted entirely
CoverageExpected pages and rows against accounted-for pages and rowsA misread digit on an accounted-for row

Keep the reason when a field cannot be accepted. An inspected blank, illegible ink, conflicting entries, an inapplicable field and a page awaiting processing call for different actions. The Text Encoding Initiative’s transcription guidelines distinguish a gap, a reading retained with uncertainty and words supplied by an editor. Preserve that distinction when a reviewer proposes a value the original cannot settle. Agreement between models that read the same faulty transcript adds no new observation of the disputed mark. TEI transcription guidelines

Quality and recovery

Measure the required fidelity

To evaluate a transcript, align it with a reviewed reading. A missing character can then count as one deletion without counting every subsequent character as wrong. Character error rate (CER) and word error rate (WER) use the same calculation at different units. Specify the character or word boundaries and apply the same normalization to both sides. Stripping punctuation before scoring prices would forgive the loss of a decimal point. JiWER alignment and transformations

edit error rate=S+D+IN\text{edit error rate}=\frac{S+D+I}{N} The alignment counts substitutions SS, deletions DD and insertions II. The denominator NN counts reference units. Extra output contributes insertions without increasing reference length, so the rate can exceed one. Specify a separate policy for an empty reference, where the fraction is undefined.
A transcript can match character for character while the exported record is unusable. Evaluate each required relation against a reference that includes it.
PropertyReference and comparison
Reading orderMatch predicted regions to reviewed regions by overlap to account for splits and merges. Compare precedence and unordered relations between corresponding regions.
Table structureGrid Table Similarity (GriTS) compares cell arrangement, content and location separately while respecting two-dimensional order.
Field valuesPrecision divides matching predictions by all predictions. Recall divides recovered required pairs by all reference pairs. Under exact field-value matching, a wrong value gives one unmatched prediction and one missed reference.
Source and item assignmentCheck the field type and localized evidence, then check the item to which the field belongs. DocILE measures localization and grouping separately from transcription and normalization.
CompletenessCompare delivered pages, regions and required items with the input inventory. Keep skipped and failed pages in the coverage denominator.

Review examples across the layouts, scripts and capture conditions expected in use. Split entire documents between training and evaluation. DocLayNet’s layout detector scored higher when other pages from the same documents appeared in training. Whole-document separation still permits a familiar template to appear on both sides, so report performance on unseen layouts separately. For an operational measure, count the units the consumer can use. In the CHICA clinic pilot, barcodes identified patient records and templates located form entries. Digit recognition accuracy was 98.6%, while 125 of 221 scanned forms could proceed without verification. The first denominator counts digits and the second counts forms. CHICA study · Evals and Benchmarks

Choose what to accept

Extraction confidence is a score for a specified prediction. Reading 40.00, associating it with a total label and getting every critical invoice field right are three different events. A column shift can spoil several associations together while leaving the words clear. Before interpreting an engine score or a model’s self-assessment as a probability, check its calibration against the event of interest on comparable labeled cases. Among cases assigned a calibrated probability near 0.9, roughly nine in ten should be correct. Machine Learning Fundamentals

An acceptance threshold divides cases between automatic release and referral. Coverage is the fraction of all eligible cases released. With correctness scored as pass or fail, selective risk is the fraction of released cases that are wrong. Risk is undefined when none are released. A stricter policy can reduce risk by referring more work. Compare policies at a common risk or coverage, taking into account both the cost of a wrong release and the available review capacity. Selective classification · Evals and Benchmarks

Basware held its extractor fixed and trained a separate confidence estimator for joint correctness of five fields: invoice number, issue date, subtotal, tax and total. The authors calibrated the estimator separately and tested on 24,000 held-out single-page, machine-readable invoices with known templates. At the production system’s undisclosed target precision, estimated automatic coverage increased from 65.6% to 73.2%. The change concerned which invoices qualified for release on that test set. Unknown templates and multipage invoices fell outside the comparison. Basware study

Make the referral reason specific. Amazon’s documented Textract review integration distinguished uncertainty about the words from uncertainty about the key-value pair. A missing critical key could trigger a search for the missing entry, while sampling regardless of confidence allowed review of apparent successes. That integration is closed to new customers. If the page itself supports competing readings, pass the candidates and their context to the reviewer. Adjusting a threshold cannot supply evidence missing from the source. Review activation conditions

Recover incomplete and disputed results

Before rerunning a job, distinguish incomplete delivery from incomplete processing. In Amazon Textract’s asynchronous API, NextToken requests another batch of available result blocks. A partial-success status and warnings about particular source pages describe a processing problem. Fetch the remaining batches before identifying what is missing. A warned page can still require investigation after the last batch arrives. GetDocumentAnalysis contract

For a disputed price, locate the OCR word selected by the field extractor. If it points to an adjacent number, revisit grouping or field assignment. If it points to the price but omits the decimal point, compare that location in the source render and the image sent to OCR. A point visible only in the render suggests changing preparation. A point clearly visible in the prepared image directs attention to recognition. If neither image contains enough detail, obtain a better capture or retain an unresolved reading. Each retry now changes an operation or supplies evidence relevant to the discrepancy. OCRmyPDF inspection artifacts

Review and correction

A reviewer needs the source version, candidate readings, referral reason and enough surrounding context to resolve the question. Figure 2 of the Vanderbilt assessment-form study places the scan beside an editable form. Selecting a field highlights the corresponding location in both views, so a reviewer can see which question a mark answers. Review time also depends on the work requested. Once transcription of handwritten comments was dropped, reported mean annotation time fell from 188 to 156 seconds, with observations above 401.5 seconds excluded from the timing analysis. Compare review policies at a fixed task scope. Vanderbilt Figure 2 and study

Record the reviewer’s decision and responsibility alongside the original reading. A reviewer may settle which item owns a value or determine that the page leaves it ambiguous. Rerun checks affected by a correction. Changing a line amount affects reconciliation, while changing a spanning header can affect every record exported under it. Keep completion and review state at page, region and field level so a local correction cannot silently mark a whole document complete. Design Engineering and AI Interfaces

Downstream use

Deliver context with the records

Export 151/597 with its row identity, governing headings and unavailable-data note. Include locations for that evidence in the processed source version, any interpretation or normalization applied, and the scope of verification. A partial-output contract can release a verified item while leaving a document-wide total pending. The converse is possible too. If every count is verified exactly once, their sum can be released even while the assignment of individual counts to recipients awaits review.

A chunk is a bounded unit supplied for retrieval. An index is a derived structure for finding such units. Docling can form chunks from the document hierarchy and repeat table headers so rows can be interpreted outside the full table. Inspect the actual chunk for qualifications and source references. Reducto’s description for lookup and HTML for reading can point to the same table. Liu’s return to the source adds a way to inspect a consequential field in context. Finding the unit and composing an answer from it are downstream responsibilities. Docling chunking · Choosing Searchable Units · Retrieval-Augmented Generation

Carry permission metadata to the consumer that enforces access. Retain lineage links from records, chunks and index entries to their source elements and extraction version. When a reviewer corrects a shared header, those links identify which independent-looking rows need to be regenerated or withdrawn. Maintain fitness through change · Privacy and Data Governance

Reference

Frequently asked questions

Does a searchable PDF still need OCR?

Check the target region. A searchable heading may sit above an image-only table, and a PDF can display correct glyphs while copying incorrect characters. Retain faithful encoded text. Recognize uncovered image text and investigate conflicts against the rendered page.

Can OCR-free extraction choose the wrong number?

Yes. An OCR-free model reads the page image directly without a separate OCR transcript. In a published example, Donut filled a publication-number field with a different number visible on the same page. To verify the field, locate the publication-number label on the page and compare its value with the extracted number.

Why can a table value need several source locations?

The cell establishes the literal value. A spanning header can supply the unit, and a separate note can qualify the reading. Retain the regions and their links so a reviewer can check which header or note applies.

Can a partially processed document supply a usable record?

Yes, when the consumer accepts that unit and all evidence required for it has been checked, including item membership. Fetch outstanding result batches and identify failed pages separately. Release a verified record with its scope explicit and withhold results that depend on missing material.

Open questions

  1. How well do document acceptance policies transfer to unfamiliar templates? A layout change can break several field associations together while preserving clear words. Measuring both automatic coverage and errors among released documents on unseen layouts and multipage inputs would show whether a threshold still supports the intended use.

  2. When is a table on the next page a continuation? Repeated headers support that interpretation but can also introduce a distinct table. Reviewed cross-page links on unfamiliar layouts, with ambiguous joins marked separately, would make it possible to measure how often an export combines unrelated records or drops continuing rows.

  3. Which referrals reduce review effort at a fixed level of residual error? Policies can appear faster by narrowing the work or referring fewer cases. Comparing time, reviewer disagreement and errors left in released records on the same forms and fields would distinguish genuine savings.

Follow the curated reading path through the speakers and demonstrations behind this entry.

19 min

AI Engineer World's Fair 2025 · 2025

Building an Agentic Platform

Ben Kus

Cited in this entry

Ben Kus explains how Box kept related fields together when dividing a large extraction request, and why its team returned to page images to check difficult readings.

Watch talk

Explore more talks

The rest of the library, beyond the curated path. Cited talks support this entry; reviewed transcripts were processed in full. Metadata candidates have not been reviewed as sources or verified as topic members.

22 matching talks

TalkSpeakerEventYear
Suman DebnathAI Engineer World's Fair 20252025
Randall HuntAI Engineer World's Fair 20252025
Stefania DrugaAI Engineer Summit 20252025
Phil NashAI Engineer Europe 20262026
Jerry LiuAI Engineer World's Fair 20252025
Samuel ColvinAI Engineer World's Fair 20252025
Nina Lopatina, Rajiv ShahAI Engineer World's Fair 20252025
Philipp SchmidAI Engineer World's Fair 20252025
Stephen Chin, Jonathan LoweAI Engineer Summit 20252025
David KaramAI Engineer World's Fair 20252025
Evaling Video Slop

Transcript reviewed

Maor BrilAI Engineer World's Fair 20262026
Denys LinkovEvent unknownUnknown
Hiral Shah, Sean SodhaEvent unknownUnknown
George HeEvent unknownUnknown
Bogdan GazaAI Engineer Code 2025Unknown
Krishna Prasad SrinivasanAI Engineer World's Fair 20262026
Palak Agarwal, Omar AlhaitAI Engineer World's Fair 20262026
Nina Žižakić, Eleonora DrykinaAI Engineer NYC 2026 Online Talks2026
Benjamin ClaviéEvent unknownUnknown
Nyah MacklinAI Engineer World's Fair 20262026
Filip MakraduliAI Engineer World's Fair 20262026
Abhi AryaEvent unknownUnknown

References

Coverage and source review
Processed transcripts
13 processed in full · 4 in the curated path
Automated source review
Passed
Metadata candidates
11 unreviewed; not verified topic membership
Corpus version
1bd8e407b26a07b33815594e1b2db5f41827119a2b3cb6fbf240f9fc571fc767

Automated review checks source support; it is not publication approval.

A synthesis of selected conference talks and technical references. Citations link to the source material; they do not imply that every talk on this subject is included.

  1. OpenCV: Geometric Image Transformations

    https://docs.opencv.org/4.13.0/da/d54/group__imgproc__transform.html

  2. PyMuPDF: Appendix 3 — Assorted Technical Information

    https://pymupdf.readthedocs.io/en/latest/app3.html

  3. Formatting strategy, Reducto slide 18 of 29 — Adit Abraham

    https://ai.engineer/talks/0I07YAuF8xA-from-ingestion-agents-ai-teams-build-on?view=slides#transcript-0I07YAuF8xA-asset-18

  4. From Ingestion to Agents: How AI Teams Build on Document Intelligence — Adit Abraham, Reducto

    https://www.youtube.com/watch?v=0I07YAuF8xA&t=667s

  5. Building the Document Context Layer for AI Agents — Jerry Liu, LlamaIndex

    https://www.youtube.com/watch?v=RQi7x-navxU&t=537s

  6. Error! The text copied from the PDF could not be pasted correctly

    https://pdfa.org/error-the-text-copied-from-the-pdf-could-not-be-pasted-correctly/

  7. Docling Technical Report

    https://arxiv.org/html/2408.09869v5

  8. PyMuPDF: OCR — Optical Character Recognition

    https://pymupdf.readthedocs.io/en/latest/recipes-ocr.html

  9. Microsoft Document Intelligence: Invoice data extraction

    https://learn.microsoft.com/en-us/azure/ai-services/document-intelligence/prebuilt/invoice?view=doc-intel-4.0.0

  10. DocILE Benchmark for Document Information Localization and Extraction

    https://arxiv.org/html/2302.05658v2

  11. From Ingestion to Agents: How AI Teams Build on Document Intelligence — Adit Abraham, Reducto

    https://www.youtube.com/watch?v=0I07YAuF8xA&t=744s

  12. Building the Document Context Layer for AI Agents — Jerry Liu, LlamaIndex

    https://www.youtube.com/watch?v=RQi7x-navxU&t=943s

  13. Forget RAG Pipelines—Build Production-Ready AI Agents in 15 Minutes

    https://www.youtube.com/watch?v=lArgRvBV3tQ&t=816

  14. Smithsonian: Farrington Character Sensing Machine 9SP3

    https://www.americanhistory.si.edu/ne/collections/object/nmah_1196719

  15. Notes on transfer of responsibility for OCR-B standards

    https://www.unicode.org/L2/L2000/00265-OCR-B.htm

  16. IBM Data Entry Marketing Guide, Z20-0841-1

    https://bitsavers.org/pdf/ibm/generalInfo/Z20-0841-1_Data_Entry_Marketing_Guide_1969.pdf

  17. Kurzweil Computer Products: Ray Kurzweil's retrospective account

    https://www.kurzweiltech.com/kcp.html

  18. An Overview of the Tesseract OCR Engine

    https://tesseract-ocr.github.io/docs/tesseracticdar2007.pdf

  19. Computer History Museum: Oral History of John Warnock, Part 2

    https://archive.computerhistory.org/resources/access/text/2023/08/102738854-05-01-acc.pdf

  20. W3C Technique PDF3: Ensuring correct tab and reading order in PDF documents

    https://www.w3.org/WAI/WCAG22/Techniques/pdf/PDF3

  21. Connectionist Temporal Classification: Labelling Unsegmented Sequence Data with Recurrent Neural Networks

    https://www.cs.toronto.edu/~graves/icml_2006.pdf

  22. High-Performance OCR for Printed English and Fraktur using LSTM Networks

    https://www.researchgate.net/publication/260341302_High-Performance_OCR_for_Printed_English_and_Fraktur_using_LSTM_Networks

  23. LayoutLM: Pre-training of Text and Layout for Document Image Understanding

    https://www.microsoft.com/en-us/research/wp-content/uploads/2020/02/layoutlm.pdf

  24. LayoutLM: Pre-training of Text and Layout for Document Image Understanding

    https://www.microsoft.com/en-us/research/wp-content/uploads/2020/02/layoutlm.pdf

  25. OCR-free Document Understanding Transformer

    https://www.ecva.net/papers/eccv_2022/papers_ECCV/papers/136880493.pdf

  26. OCRmyPDF: Advanced features

    https://ocrmypdf.readthedocs.io/en/latest/advanced.html

  27. OCRmyPDF: v17 release notes

    https://ocrmypdf.readthedocs.io/en/latest/releasenotes/version17.html

  28. Microsoft: Structure of a WordprocessingML document

    https://learn.microsoft.com/en-us/office/open-xml/word/structure-of-a-wordprocessingml-document

  29. PROV-DM: The PROV Data Model

    https://www.w3.org/TR/prov-dm/

  30. Amazon Textract: GetDocumentAnalysis

    https://docs.aws.amazon.com/textract/latest/APIReference/API_GetDocumentAnalysis.html

  31. Docling: Docling document

    https://docling-project.github.io/docling/concepts/docling_document/

  32. Structuring the Unstructured: Advanced Document Parsing for AI Workflows

    https://www.youtube.com/watch?v=-x5GEVnkuRw&t=714

  33. Structuring the Unstructured: Advanced Document Parsing for AI Workflows

    https://www.youtube.com/watch?v=-x5GEVnkuRw&t=804

  34. hOCR 1.2: OCR Workflow and Output embedded in HTML

    https://kba.github.io/hocr-spec/1.2/

  35. W3C: Web Annotation Data Model

    https://www.w3.org/TR/annotation-model/

  36. Library of Congress: ALTO — Technical Metadata for Layout and Text Objects

    https://www.loc.gov/standards/alto/

  37. 2017 PAGE XML Format for Page Content

    https://www.primaresearch.org/schema/PAGE/gts/pagecontent/2017-07-15/Simple%20PAGE%20XML%20Example.pdf

  38. Google Document AI: Document REST resource

    https://docs.cloud.google.com/document-ai/docs/reference/rest/v1/Document

  39. Tesseract: Improving the quality of the output

    https://tesseract-ocr.github.io/tessdoc/ImproveQuality.html

  40. DewarpNet: Official implementation and OCR evaluation notes

    https://github.com/cvlab-stonybrook/DewarpNet

  41. OCRmyPDF: Advanced features — Debugging the intermediate files

    https://ocrmypdf.readthedocs.io/en/latest/advanced.html

  42. PyMuPDF: Page

    https://pymupdf.readthedocs.io/en/latest/page.html

  43. Hugging Face Evaluate: Character Error Rate

    https://huggingface.co/spaces/evaluate-metric/cer/blob/main/README.md

  44. An End-to-End Trainable Neural Network for Image-based Sequence Recognition and Its Application to Scene Text Recognition

    https://arxiv.org/pdf/1507.05717

  45. Connectionist Temporal Classification: Labelling Unsegmented Sequence Data with Recurrent Neural Networks

    https://www.cs.toronto.edu/~graves/icml_2006.pdf

  46. Tesseract FAQ

    https://tesseract-ocr.github.io/tessdoc/FAQ.html

  47. Building an Agentic Platform

    https://www.youtube.com/watch?v=12v5S1n1eOY&t=438

  48. NISO JATS: Table Wrapper Footer

    https://jats.nlm.nih.gov/publishing/tag-library/1.1d2/element/table-wrap-foot.html

  49. Recursive X-Y Cut using Bounding Boxes of Connected Components

    https://haralick.org/conferences/71280952.pdf

  50. Empirical Performance Evaluation Methodology and Its Application to Page Segmentation Algorithms

    https://lhncbc.nlm.nih.gov/LHC-publications/PDF/pub2001008.pdf

  51. DocLayNet: A Large Human-Annotated Dataset for Document-Layout Analysis

    https://arxiv.org/pdf/2206.01062

  52. The Significance of Reading Order in Document Recognition and Its Evaluation

    https://primaresearch.org/www/assets/papers/ICDAR2013_Clausner_ReadingOrder.pdf

  53. Unicode Standard Annex #9: Unicode Bidirectional Algorithm, Revision 51

    https://www.unicode.org/reports/tr9/tr9-51.html

  54. Tagged PDF Best Practice Guide: Syntax 1.0.1

    https://pdfa.org/download-area/publications/Tagged-PDF-Best-Practice-Guide.pdf

  55. ICU4J: Edits.Iterator

    https://unicode-org.github.io/icu-docs/apidoc/released/icu4j/com/ibm/icu/text/Edits.Iterator.html

  56. PyMuPDF: Document

    https://pymupdf.readthedocs.io/en/latest/document.html

  57. Amazon Textract: Text Detection and Document Analysis Response Objects

    https://docs.aws.amazon.com/textract/latest/dg/how-it-works-document-layout.html

  58. Structuring the Unstructured: Advanced Document Parsing for AI Workflows — Cedric Clyburn

    https://www.youtube.com/watch?v=-x5GEVnkuRw&t=612s

  59. Structuring the Unstructured: Advanced Document Parsing for AI Workflows

    https://www.youtube.com/watch?v=-x5GEVnkuRw&t=1179

  60. LayoutLMv2: Multi-modal Pre-training for Visually-rich Document Understanding

    https://aclanthology.org/2021.acl-long.201.pdf

  61. OCR-free Document Understanding Transformer

    https://arxiv.org/pdf/2111.15664

  62. IBM Granite-Docling-258M model card

    https://huggingface.co/ibm-granite/granite-docling-258M

  63. Gemini API: Document understanding

    https://ai.google.dev/gemini-api/docs/document-processing

  64. AI Engineering with the Google Gemini 2.5 Model Family

    https://www.youtube.com/watch?v=zK9lYrLbjSg&t=2580

  65. W3C WAI: Tables with Multi-Level Headers

    https://www.w3.org/WAI/tutorials/tables/multi-level/

  66. Microsoft Table Transformer: Official implementation

    https://github.com/microsoft/table-transformer

  67. PubTables-1M: Towards comprehensive table extraction from unstructured documents

    https://arxiv.org/html/2110.00061v3

  68. Microsoft Document Intelligence: Custom template document model

    https://learn.microsoft.com/en-us/azure/ai-services/document-intelligence/train/custom-template?view=doc-intel-4.0.0

  69. Amazon Textract: Selection Elements

    https://docs.aws.amazon.com/textract/latest/dg/how-it-works-selectables.html

  70. Building an Agentic Platform

    https://www.youtube.com/watch?v=12v5S1n1eOY&t=473

  71. Google Document AI samples: web-app-pix2info-python

    https://github.com/GoogleCloudPlatform/document-ai-samples/blob/main/web-app-pix2info-python/README.md

  72. TEI P4 Guidelines: Transcription of Primary Sources

    https://tei-c.org/Vault/P4/doc/pdf/PH.pdf

  73. Extracting Medical Information from Paper COVID-19 Assessment Forms

    https://www.thieme-connect.com/products/ejournals/pdf/10.1055/s-0041-1723024.pdf

  74. Peppol BIS Billing 3.0: Rule BR-CO-13

    https://docs.peppol.eu/poacc/billing/3.0/rules/ubl-tc434/BR-CO-13/

  75. OWASP Input Validation Cheat Sheet

    https://cheatsheetseries.owasp.org/cheatsheets/Input_Validation_Cheat_Sheet.html

  76. OWASP Input Validation Cheat Sheet

    https://cheatsheetseries.owasp.org/cheatsheets/Input_Validation_Cheat_Sheet.html

  77. Building an Agentic Platform

    https://www.youtube.com/watch?v=12v5S1n1eOY&t=707

  78. Hugging Face Evaluate: Word Error Rate

    https://huggingface.co/spaces/evaluate-metric/wer/blob/main/README.md

  79. JiWER: Text Transformations and Alignment

    https://jitsi.github.io/jiwer/usage/

  80. Introduction to Information Retrieval: Precision and recall

    https://nlp.stanford.edu/IR-book/html/htmledition/evaluation-of-unranked-retrieval-sets-1.html

  81. Using Adaptive Turnaround Documents to Electronically Acquire Structured Data in Clinical Settings

    https://scholarworks.indianapolis.iu.edu/bitstreams/afbc7415-89cc-45c3-a620-68d4581a3381/download

  82. Amazon SageMaker AI: Human Loop Activation Conditions for Amazon Textract

    https://docs.aws.amazon.com/sagemaker/latest/dg/a2i-json-humantaskactivationconditions-textract-example.html

  83. On Calibration of Modern Neural Networks

    https://proceedings.mlr.press/v70/guo17a/guo17a.pdf

  84. Failure Prediction in 2D Document Information Extraction with Calibrated Confidence Scores

    https://helda.helsinki.fi/server/api/core/bitstreams/66f035d7-78bc-42bb-b6be-912895dfbf06/content

  85. Selective Classification: Coverage and Conditional Risk

    https://arxiv.org/pdf/1705.08500

  86. Forget RAG Pipelines—Build Production-Ready AI Agents in 15 Minutes

    https://www.youtube.com/watch?v=lArgRvBV3tQ&t=1892

  87. Docling: Chunking

    https://docling-project.github.io/docling/concepts/chunking/

  88. Building AI Agents that actually automate Knowledge Work

    https://www.youtube.com/watch?v=jVGCulhBRZI&t=254

  89. Introduction to Information Retrieval: Near-duplicates and shingling

    https://nlp.stanford.edu/IR-book/html/htmledition/near-duplicates-and-shingling-1.html

  90. Structuring the Unstructured: Advanced Document Parsing for AI Workflows

    https://www.youtube.com/watch?v=-x5GEVnkuRw&t=176

  91. Forget RAG Pipelines—Build Production-Ready AI Agents in 15 Minutes

    https://www.youtube.com/watch?v=lArgRvBV3tQ&t=1971