Define your labels, ask for content-based category suggestions, and keep uncertain documents visible for review.
AI document classification assigns categories to documents based on their content and a defined set of labels. A classification task may group sources by topic, purpose, or document type. A useful result explains why a label fits and makes uncertain cases visible instead of forcing every file into a category.
Use iWeaver’s document conversation to ask for category suggestions against rules you provide. Start with a small, focused collection, define what each label includes, and review the proposed assignments. This is a way to plan document organization; suggested labels do not automatically rename files, move folders, or set access permissions.
Decide whether you are sorting by topic, document type, or workflow purpose. Mixing these dimensions can produce confusing labels: a research paper may also be background reading. Classify each dimension separately when you need both.
Give each label an inclusion rule and, when helpful, an exclusion example. “Interview notes” and “published research” are easier to apply than a broad label such as “useful information.” Clear definitions make the output easier for a person to review.
Request a short explanation and a supporting passage or section when available. A filename alone may be misleading, so tell the AI to consider the provided content. Inspect whether the stated reason actually satisfies your category rule.
Specify whether more than one label is allowed. Provide a review category for documents that do not clearly fit, and ask for the ambiguity to be explained. You can then refine the rules without hiding difficult cases in a miscellaneous folder.
Imagine a project pack containing papers, interview notes, and draft recommendations. The example below sorts by a document’s role in the project. It is a proposed taxonomy you can adapt, not a claim that the files have already been classified.
| Category | Include when | Keep separate from |
|---|---|---|
| Background research | The document supplies existing evidence or context | A new recommendation written for this project |
| Interview material | The document records a conversation or interview | A later summary combining several sources |
| Working deliverable | The document develops a project-specific output | An external source used to support it |
| Needs review | The role is unclear or more than one rule conflicts | A forced label without a supported reason |
Try: “Suggest a category for each source using these definitions. Return source name, proposed label, reason, and review note. Do not force a label when the content is insufficient. Classify document purpose separately from subject topic.”
Write a short definition for each category and explain whether labels are exclusive. Add an unresolved category and describe how overlapping content should be handled. Start with one classification dimension, such as topic or purpose.
Choose a focused set of documents and identify each source clearly. Ask for source name, suggested label, a content-based reason, and any ambiguity. Treat the response as a proposal to inspect, rather than an automatic filing action.
Compare proposed labels with your definitions and inspect ambiguous sources. If the same disagreement keeps appearing, improve the rule before expanding the collection. Keep the original documents and record any reviewed decisions you need to reuse.
Define the themes relevant to your research question and ask for proposed topic labels. Allow more than one theme when a paper spans several subjects. Check the reason for each suggestion before deciding how to organize your reading.
Organize a Research Reading List
Use purpose-based categories to distinguish background research, interview material, and working outputs. Ask for the reason a file belongs in each group. Review blended documents carefully so a draft recommendation is not mistaken for an independent source.
Structure Project Categories
Separate explanations, worked examples, and practice questions using rules that match your study plan. A document may support more than one activity, so decide whether multiple labels are useful. Use the reviewed groups to plan the next study session.
Organize Learning Resources
Define themes based on the questions your audience asks. Request proposed labels and a note where a source fits several ideas. Review the groups before turning them into an editorial outline or a set of future topics.
Group Content SourcesA file may contain several document types or cover more than one topic. A single label can lose that distinction. Decide whether to permit several labels or review the relevant sections individually; do not assume that a category suggestion splits the file.
A topic or purpose label helps arrange material for the next task. It should not silently become a quality score, confidentiality decision, or access-control rule. Those decisions need their own definitions and an appropriate review process.
Choose a focused tool for the next part of your document workflow.
Understand what a source covers before deciding where it belongs.
Summarize a Source →Explore the common themes in a focused set of related documents.
Explore Shared Themes →Visualize relationships after you have reviewed your category structure.
Map the Categories →Bring a focused document set and a short list of labels. Ask for reasoned suggestions, review the uncertain cases, and refine the structure.
Start Organizing Categories