MMaterials Literature Database
Expanding

LITERATURE → STRUCTURED EVIDENCE

Turn literature into
a searchable materials space.

Full-text extraction results from different materials domains, brought together in one place to explore composition, synthesis, characterization, reaction conditions, and performance evidence. Every record retains its literature source and original reporting basis.

Build your own database →
—structured papers
—materials / catalysts
—performance metrics
—research domains
100%traceable records

DATASET REGISTRY

Starting with one domain, growing into a materials database.

Each dataset has its own schema, version, and quality notes, and joins the site through a shared registry.

＋
NEXT

EXTENSIBLE BY DESIGN

Next materials domain

Once its schema has been defined, extracted, and validated, a new domain can be added as an independent dataset.

Dataset registry

EVIDENCE EXPLORER

Structured evidence explorer

Version —

—

Overview

0 / 0 recordsSelect any row to inspect every field and its original reporting basis
No matching records. Try relaxing the filters.

DATABASE-GROUNDED SCIENTIFIC DIALOGUE

Discuss the database—even when
a question has no single right answer.

The model first searches structured records and evidence fields, then compares papers, develops mechanistic inferences, and proposes experiments. Facts, inferences, and hypotheses must be stated separately.

01 Database facts must be cited02 Cross-paper conclusions are labeled as inferences03 Experimental ideas include a validation route
DeepSeek V4 Flash · Thinking
DISCUSSION STARTERS

Start with a falsifiable question.

Enter to send · Shift + Enter for a new line · Answers may contain inferences; verify critical decisions against the source papers

EVIDENCE-PRESERVING WORKFLOW

Not merely a summary,
but extraction that preserves context.

Material names, capacity bases, and experimental conditions are not forcibly flattened across papers. The site presents standardized fields while retaining reported wording, evidence locations, and uncertainty.

  1. 01
    Domain schema

    Define the granularity of material, state, device, and performance records first.

  2. 02
    Full-text structured extraction

    Extract by section and keep pristine, reacted, and intermediate samples distinct.

  3. 03
    Validation and database construction

    After schema validation, build tables, evidence indexes, and the database.

  4. 04
    Versioned publication

    Add each dataset through the registry while retaining provenance and version information.