# How BisQue organizes your work

> Understand the objects that connect data, annotation, and analysis in BisQue.

BisQue preserves scientific context by treating data, organization, and analysis as related
objects instead of loose files.

```text
Resources ──► Datasets
    │             │
    └────► Projects ──► Annotation branches
               │
               └────► Workflow revisions ──► Runs ──► Artifacts
```

## Resources are the foundation

A **resource** is an item managed by BisQue, such as an image, file, table, dataset, model, or
workflow. Resources can carry hierarchical tags and access rules. Specialized viewers make
the underlying content useful without hiding its resource identity.

## Datasets collect resources

A **dataset** is an ordered, one-level collection of canonical non-dataset resources. Use
datasets to group inputs that should travel together without copying the underlying data;
datasets themselves cannot be members of another dataset.

## Projects are collaborative workspaces

A **project** brings resources together for a defined annotation or analysis task. It owns its
membership, label configuration, cards, annotation branches, and saved workflows. Project
roles determine who may view, contribute, or administer the work.

## Workflows make analysis repeatable

A **workflow** is a directed graph of module steps and inputs. Saving creates an immutable
revision. A run records the exact revision, resolved module versions, authorized inputs, and
produced artifacts so results can be traced back to their origin.

## Provenance connects results to inputs

BisQue records derived outputs as resources and artifacts rather than anonymous downloads.
This makes it possible to inspect what ran, which data it used, and how an output relates to
the project that produced it.