Unstructured data
Unstructured data is information that lacks a predefined schema or consistently organized fields.
Elasticsearch in Action, Second Edition
Unstructured data is data that does not conform to a fixed field-based or tabular schema, such as free-form text.
- How it is used
- Supports full-text search when users need relevant results from free-form text instead of exact field matches.
- How deep it goes
- Defines schema-free data and explains relevance-oriented querying, then defers analyzer configuration for synonyms, stemming, and spelling variations.
- Examples
- Uses a quotation about imagination and knowledge to contrast direct word matches with synonyms, misspellings, and abbreviations that miss.
- What you can do after
- Apply it via Elasticsearch full-text queries to retrieve and rank documents matching users’ text searches.
Read how Elasticsearch in Action, Second Edition explains unstructured data →
AI-Powered Search
Unstructured data is information without a rigid predefined schema, often containing rich content and relationships in flexible forms such as text, images, video, and audio.
- How it is used
- Frames search content as flexible information whose hidden relationships can help answer queries.
- How deep it goes
- Defines the term, compares document links with SQL joins, and works through entity resolution; defers hands-on implementation.
- Examples
- Uses text, media, event logs, conference documents, and differently written person names to expose structure and ambiguity.
- What you can do after
- Evaluate/understand how search documents can be related without explicit foreign keys and why ambiguous matches need care.
Data Without Labels
Unstructured data lacks an easily usable row-and-column schema and may include text, images, audio, or video.
- How it is used
- Guides the choice of preparation and analysis for text, images, audio, and video that do not readily form tables.
- How deep it goes
- Defines the category and explains numerical image pixels and text vectors, but does not specify one conversion procedure for every modality.
- Examples
- Uses customer reviews and transcribed calls as text sources, alongside medical-image and autonomous-driving applications.
- What you can do after
- Evaluate/understand it by recognizing when data needs numerical representation and potentially more advanced analysis than a structured table.
Delivering Machine Learning Projects: From design to deployment
An unstructured field is a data field containing content such as images, text, or sounds without a fixed tabular representation.
- How it is used
- Guides early checks of whether image, text, and other unstructured resources are useful and what quality risks need investigation.
- How deep it goes
- Provides survey checks and image-EDA criteria, while deferring involved investigation beyond the initial survey.
- Examples
- Uses building incident descriptions and sampled images to illustrate useful content, missing values, repetition, and labeling risks.
- What you can do after
- Practice/perform a preliminary unstructured-data survey and identify image-quality questions for deeper exploration.
Essential GraphRAG
Unstructured data is information without a predefined tabular or schema-based organization, such as free-form text.
- How it is used
- Supports contextual retrieval in RAG while motivating conversion into structured representations for precise questions.
- How deep it goes
- Explains retrieval limits and extraction into graphs, but stops short of detailing extraction prompts or database schemas.
- Examples
- Uses article text, legal contracts, customer-support transcripts, and medical information as representative unstructured sources.
- What you can do after
- Build it from scratch by extracting structured information from raw text and importing it into a knowledge graph.
Machine Learning for Tabular Data: XGBoost, Deep Learning, and AI
Data whose varied modalities and organization do not conform to a fixed rows-and-columns schema.
- How it is used
- Frames the practical choice between tabular workflows and deep-learning approaches for varied, non-matrix data.
- How deep it goes
- Defines the format, lists modalities, explains feature engineering, and reaches end-to-end representation learning without implementation details.
- Examples
- Uses images, video, audio, text, JSON sensor data, and social media streams as representative inputs.
- What you can do after
- Evaluate whether a dataset requires handcrafted feature extraction or can support end-to-end deep-learning transformations.
Fast Python: High performance techniques for large datasets
- How it is used
- Frames unstructured data as a growing source of computational demand that motivates efficient data processing.
- How deep it goes
- Connects unstructured representation to greater processing demands, while deferring detailed treatment.
Read how Fast Python: High performance techniques for large datasets explains unstructured data →
The Art of AI Product Development
Unstructured data is information without a consistent predefined tabular or schema-based organization, such as free text or images.
- How it is used
- Guides the preparation and retrieval of text, images, calls, and emails for AI training, inference, and product analysis.
- How deep it goes
- Defines the category and identifies preprocessing needs, but stops short of detailing specific cleaning or structuring procedures.
- Examples
- Uses free text, images, sales calls, emails, user interviews, and product reviews as representative sources.
- What you can do after
- Apply it via retrieval-augmented generation and data-engineering pipelines while separating unstructured from structured sources.
Read how The Art of AI Product Development explains unstructured data →
Machine Learning System Design
Data that lacks a fixed machine-readable schema for its content, such as free text or images.
- How it is used
- Positions unstructured data as multimodal input for deep-learning systems, feature extraction, monitoring, and similarity retrieval.
- How deep it goes
- Explains embeddings, centralized storage, derived monitoring signals, and staged validation, without detailing model internals.
- Examples
- Uses marketplace images, descriptions, reviews, image brightness, text length, and audio-frequency distributions as concrete cases.
- What you can do after
- Diagnose unstructured-data failures with derived statistics and incorporate embeddings into shared ML pipelines when justified.
Read how Machine Learning System Design explains unstructured data →
Knowledge Graphs and LLMs in Action
Unstructured data is information lacking a fixed tabular or schema-constrained organization, such as free-form documents.
- How it is used
- Feeds knowledge-graph construction, retrieval, recommendation, decision support, and other applications that require mining distributed information.
- How deep it goes
- Explains the extraction pipeline and linguistic obstacles, but stops short of detailed algorithmic or implementation specifications.
- Examples
- Uses emails, legislation, research papers, regulatory reports, clinical documents, and scientific publications as source artifacts.
- What you can do after
- Build knowledge graphs from documents by extracting entities and relationships, then support retrieval, integration, and hypothesis generation.
Read how Knowledge Graphs and LLMs in Action explains unstructured data →
Financial AI in Practice
Information that does not conform to a predefined schema for direct machine-readable interpretation.
- How it is used
- Frames unstructured data as one of the diverse inputs analyzed in financial AI and fraud detection.
- How deep it goes
- Names financial examples and describes multisource integration, but gives no schema, representation, or processing procedure.
- Examples
- Identifies regulatory reports, filings, and customer communications as financial examples.
Read how Financial AI in Practice explains unstructured data →
Related concepts
- structured datacontrasts with · Structured data is information organized according to an explicit schema or predictable fields, types, records, or relationships for consistent machine processing.
- textincludes · Text is written or digitally encoded linguistic content represented as characters, words, tokens, or other textual units.
- retrieval-augmented generationused by · Retrieval-augmented generation is a technique that retrieves relevant information from external sources and supplies it as context to a generative model when producing an output.
- language modelused by · A language model is a computational model that represents patterns in linguistic tokens to estimate sequence probabilities and predict or generate language from context.
- deep learningused by · Deep learning is a machine-learning approach that uses neural networks with multiple computational layers to learn representations from data.
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.