Structured extraction
Structured extraction is the process of identifying specified information in source content and representing it as organized data in a defined structure.
Learn AI Data Engineering
AI-powered extraction is the use of an artificial intelligence model to identify and retrieve requested fields from source content based on field descriptions rather than fixed location rules.
- How it is used
- Transforms variable JSON, HTML, and free-form text into records usable by downstream analytics and pipeline stages.
- How deep it goes
- Works through schemas, prompts, typed outputs, validation, batch evaluation, cost tracking, and narrowly scoped retries.
- Examples
- Uses news-article JSON, product pages across brands, and baseball play-by-play descriptions as extraction inputs.
- What you can do after
- Apply it via OpenAI and Pydantic to extract schema-shaped fields, evaluate coverage, and operate batch workflows with failure tracking.
Read how Learn AI Data Engineering explains structured extraction →
Essential GraphRAG
Structured data extraction is the process of converting information from unstructured sources into machine-readable data organized by explicit fields and relationships.
- How it is used
- Uses LLMs to convert documents into structured records for filtering, aggregation, retrieval, and knowledge graph construction.
- How deep it goes
- Explains schema design, prompting guidance, nested objects, enums, graph import, and clause-based chunking without detailing extraction evaluation.
- Examples
- Demonstrates calendar events, legal contracts, organizations, locations, JSON records, enums, and clause-based document chunks.
- What you can do after
- Build it from scratch by defining schemas, prompting LLMs, extracting records, and importing results into a graph database.
Read how Essential GraphRAG explains structured extraction →
Natural Language Processing in Action, Second Edition
Information extraction is the automatic identification and structuring of facts, entities, and relations from unstructured text.
- How it is used
- Supports factual question answering, knowledge-graph construction, chatbot actions, and grounding for generated language.
- How deep it goes
- Combines patterns, named-entity recognition, coreference resolution, relation extraction, tagging, and parsing, without detailing model internals.
- Examples
- Demonstrates reminder interpretation and extraction experiments on a Wikipedia article about Timnit Gebru.
- What you can do after
- Build it with spaCy components, combine extracted results, and populate or query a knowledge graph.
Read how Natural Language Processing in Action, Second Edition explains structured extraction →
Data Analysis with LLMs
Information extraction is the task of identifying and structuring specified facts or entities from text.
- How it is used
- Turns specified facts in text into structured attributes for tasks such as comparing applicants or compiling biographical data.
- How deep it goes
- Details prompt components and parsing rows and fields, but does not describe model internals.
- Examples
- Uses applicant emails to extract names, GPAs, and degrees, and biographies to extract birthplaces and dates.
- What you can do after
- Build it from scratch by prompting a language model and parsing its response into rows of structured data.
Read how Data Analysis with LLMs explains structured extraction →
D3.js in Action, Third Edition
Data extraction is the process of obtaining selected information from a source such as a web page or document.
- How it is used
- Supports custom visualizations by converting selected web-page content into structured fields.
- How deep it goes
- Demonstrates table classification, DOM selection, row iteration, and field collection through a browser-console script.
- Examples
- Uses Van Gogh painting tables from Wikipedia and extracts titles, image URLs, and mediums for a timeline.
- What you can do after
- Build it from scratch by extracting selected records from HTML tables in a browser console.
Read how D3.js in Action, Third Edition explains structured extraction →
LLM Customization and Fine-Tuning
The process of identifying and extracting specified fields from unstructured or semi-structured sources.
- How it is used
- Uses many-shot learning to cover varied input formats in structured extraction tasks.
- How deep it goes
- Explains why layout coverage matters and stops short of specifying an extraction implementation.
- Examples
- Uses invoices from vendors with varied layouts to illustrate the need for representative examples.
- What you can do after
- Apply many-shot prompting by including at least one example for each input layout family.
Read how LLM Customization and Fine-Tuning explains structured extraction →
Also covered in
- The Well-Grounded Data Analyst
- Fight Fraud with Machine LearningIn early access (MEAP)
- Graph Algorithms for Data Science
- ScyllaDB in Action
- Build a Website with ChatGPT
- Knowledge Graphs and LLMs in Action
- AI for Everyday IT
- Build an AI Agent (From Scratch)
- AI Agents and Applications
- Snowflake Data Engineering
- AI-Powered Search
- The Quick Python Book, Fourth Edition
What to know first
- language modelA language model is a computational model that represents patterns in linguistic sequences to assign probabilities, predict, generate, or evaluate language.
- JSONJSON is a language-independent text-based data-interchange format that represents structured values using objects, arrays, strings, numbers, booleans, and null.
- knowledge graphA knowledge graph is a graph-structured representation of entities and their semantically meaningful relationships, often including associated attributes or facts.
- ChatGPTChatGPT is OpenAI’s conversational AI service, which lets users interact with generative AI models and related tools through natural-language prompts.
Related concepts
- language modeluses · A language model is a computational model that represents patterns in linguistic sequences to assign probabilities, predict, generate, or evaluate language.
- structured dataproduces · Structured data is information organized according to a defined schema or arrangement that makes its elements and relationships consistently interpretable and processable.
- dependency graphuses · A dependency graph is a graph whose nodes represent entities and whose edges represent dependency relationships among them.
- regular expressionused by · A regular expression is a formal pattern notation that describes sets of strings and specifies how text can be matched, searched, or transformed.
- coreference resolutionhas part · Coreference resolution is the task of identifying expressions in text that refer to the same entity and linking those expressions to it.
- data sourceuses · A data source is an origin, system, repository, or process from which data is obtained or made available for use or processing.
Questions about structured extraction
What is structured extraction?
Structured extraction is the process of identifying specified information in source content and representing it as organized data in a defined structure.
Which Manning books explain structured extraction?
Learn AI Data Engineering, Essential GraphRAG, Natural Language Processing in Action, Second Edition, Data Analysis with LLMs, D3.js in Action, Third Edition and LLM Customization and Fine-Tuning.
What should I know before learning structured extraction?
Manning books that teach structured extraction usually introduce language model, JSON, knowledge graph and ChatGPT first.
What are the parts of structured extraction?
Structured extraction includes coreference resolution.
What is structured extraction also called?
Structured extraction is also called structured data extraction.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.





