LayoutLM
E435880
LayoutLM is a transformer-based document understanding model that jointly leverages text, layout, and visual information to process and analyze scanned documents and forms.
All labels observed (4)
| Label | Occurrences |
|---|---|
| LayoutLM canonical | 1 |
| LayoutLM: Pre-training of Text and Layout for Document Image Understanding | 1 |
| LayoutLMv2 | 1 |
| LayoutLMv3 | 1 |
How this entity was disambiguated
This entity first appeared as the object of triple T4389208 — resolving that mention is where its identity was fixed. The disambiguator weighed these candidate entities and picked the highlighted one (or “None”, minting a new entity). This is how homonymy is resolved: the same surface form can point to different entities.
NED1
Entity disambiguation (via context triple)
gpt-5-mini-2025-08-07
Target entity: LayoutLM Context triple: [Hugging Face Transformers, supportsModelType, LayoutLM]
-
A.
HOCR
HOCR is the commonly used abbreviation for the Head of the Charles Regatta, a major annual rowing event held on the Charles River in Boston and Cambridge, Massachusetts.
-
B.
OCR
OCR is the Office for Civil Rights, a U.S. government agency responsible for enforcing civil rights laws and ensuring equal access and non-discrimination in federally funded programs.
-
C.
Kurzweil OCR (optical character recognition) systems
Kurzweil OCR (optical character recognition) systems are pioneering software tools that convert printed text into digital, machine-readable form, widely used for document digitization and accessibility for the visually impaired.
-
D.
Gradient-based learning applied to document recognition
"Gradient-based learning applied to document recognition" is a seminal 1998 paper by Yann LeCun and colleagues that introduced and demonstrated the effectiveness of convolutional neural networks for tasks like handwritten digit recognition, helping to lay the foundations of modern deep learning.
-
E.
DocSend
DocSend is a document-sharing and tracking platform designed for securely sending files and gaining analytics on how recipients engage with them.
- F. None of above. chosen
- G. Unsure - the case is ambiguous/there is not enough information to decide.
NED2
Entity disambiguation (via description)
gpt-5-mini-2025-08-07
Target entity: LayoutLM Target entity description: LayoutLM is a transformer-based document understanding model that jointly leverages text, layout, and visual information to process and analyze scanned documents and forms.
-
A.
HOCR
HOCR is the commonly used abbreviation for the Head of the Charles Regatta, a major annual rowing event held on the Charles River in Boston and Cambridge, Massachusetts.
-
B.
OCR
OCR is the Office for Civil Rights, a U.S. government agency responsible for enforcing civil rights laws and ensuring equal access and non-discrimination in federally funded programs.
-
C.
Kurzweil OCR (optical character recognition) systems
Kurzweil OCR (optical character recognition) systems are pioneering software tools that convert printed text into digital, machine-readable form, widely used for document digitization and accessibility for the visually impaired.
-
D.
Gradient-based learning applied to document recognition
"Gradient-based learning applied to document recognition" is a seminal 1998 paper by Yann LeCun and colleagues that introduced and demonstrated the effectiveness of convolutional neural networks for tasks like handwritten digit recognition, helping to lay the foundations of modern deep learning.
-
E.
DocSend
DocSend is a document-sharing and tracking platform designed for securely sending files and gaining analytics on how recipients engage with them.
- F. None of above. chosen
Statements (45)
| Predicate | Object |
|---|---|
| instanceOf |
document understanding model
ⓘ
multimodal transformer model ⓘ pretrained language model ⓘ |
| availableAs | open-source model ⓘ |
| basedOn | Transformer architecture ⓘ |
| category |
document AI model
ⓘ
vision-language model ⓘ |
| designedFor |
document image classification
ⓘ
document understanding ⓘ form understanding ⓘ forms ⓘ information extraction ⓘ key information extraction ⓘ scanned documents ⓘ |
| developer | Microsoft Research Asia ⓘ |
| hasAuthor |
Furu Wei
ⓘ
Lei Cui ⓘ Ming Zhou ⓘ Minghao Li ⓘ
linked to:
Yiheng Xu
Shaohan Huang ⓘ Yiheng Xu ⓘ |
| hasVersion |
LayoutLMv2
ⓘ
linked to:
LayoutLM
LayoutLMv3 ⓘ
linked to:
LayoutLM
|
| hostedOn | Hugging Face Transformers ⓘ |
| implementedIn | PyTorch ⓘ |
| inputModality |
image
ⓘ
layout ⓘ text ⓘ |
| introducedAt |
KDD 2020
ⓘ
linked to:
SIGKDD
|
| introducedIn | 2019 ⓘ |
| language | English ⓘ |
| leverages |
layout information
ⓘ
text information ⓘ visual information ⓘ |
| optimizationObjective |
masked language modeling
ⓘ
multi-task learning for document understanding ⓘ |
| paperTitle |
LayoutLM: Pre-training of Text and Layout for Document Image Understanding
ⓘ
linked to:
LayoutLM
|
| pretrainedOn | large-scale document image datasets ⓘ |
| supportsTask |
document question answering
ⓘ
invoice understanding ⓘ receipt understanding ⓘ |
| uses |
2D positional embeddings
ⓘ
bounding box coordinates ⓘ image region features ⓘ token-level text embeddings ⓘ |
How these facts were elicited
The pipeline generated the facts above by prompting gpt-5.1 with this entity's name + description and the instruction below.
Instruction
You are a knowledge base construction expert. Given a subject entity and a description of it, return factual statements that you know for the subject as a JSON list of dictionaries(triples), where keys must be "subject", "predicate" and "object". The number of facts may be very high, between 25 to 50 or more, for very popular subjects. For less popular subjects, the number of facts can be very low, like 5 or 10. # Requirements - If you don't know the subject at all, return an empty list. - If the subject is not a named entity, return an empty list. - Include at least one triple where predicate is "instanceOf". - Do not get too wordy. - Separate several objects into multiple triples with one object.
Input
Subject: LayoutLM Description of subject: LayoutLM is a transformer-based document understanding model that jointly leverages text, layout, and visual information to process and analyze scanned documents and forms.
Referenced by (4)
Full triples — surface form annotated when it differs from this entity's canonical label.
linked to: LayoutLM
linked to: LayoutLM
LayoutLM
→
paperTitle
→
LayoutLM: Pre-training of Text and Layout for Document Image Understanding
ⓘ
linked to: LayoutLM