Releases: XiaoheMo9178/paper-classifier
Release list
Paper Classifier v1.2.0
Paper Classifier v1.2.0 Release Notes
Release type: Workflow and taxonomy update
Plugin IDs: paper-classifier@example.com, paper-classifier-en@example.com
What Changed
- Added a research topic prompt before classification.
- The plugin now classifies papers by their role in the user-entered research topic, rather than only by each paper's own study type.
- The last entered research topic is saved and prefilled for the next classification run.
- Updated the controlled taxonomy to research-topic-oriented buckets:
- Core Topic Research
- Background Theory and Concepts
- Methods Models and Tools
- Measurement Evaluation and Indicators
- Intervention Application and Practice
- Mechanisms and Risk Factors
- Evidence Synthesis and Review
- Data Resources and Systems
- Policy Ethics and Translation
- Weakly Related or Exclude
Result
When you classify a batch, the workflow is now:
Select papers -> AI Classify Papers -> enter research topic -> classify by role in that topic
This should make the final collection tree reflect the user's actual project topic instead of only generic paper types.
Paper Classifier v1.0.0
Paper Classifier v1.0.0 — Release Notes
Release Date: 2026-02-24
Plugin ID: paper-classifier@example.com
Compatibility: Zotero 7 / 8 (Bootstrap extension model)
What's New in v1.0.0
This is the first public release of Paper Classifier, a Zotero plugin that uses the DeepSeek AI API to automatically classify academic papers by research theme and organize them into a hierarchical collection tree.
Key Features
AI-Powered Research Theme Classification
Papers are classified by research theme rather than academic discipline. The AI model identifies themes such as Randomized Controlled Trial, Systematic Review, Meta-Analysis, Mechanistic Study, Scale Development & Validation, Intervention Study, and more — or creates appropriate custom theme names when needed.
Classification output format: Primary Theme / Secondary Theme
Automatic Three-Level Collection Hierarchy
The plugin creates and maintains a three-level collection structure in your Zotero library automatically:
<Collection Root> (default: "AI主题分类")
└── <Primary Theme>
└── <Secondary Theme>
└── [paper items]
Missing intermediate collections are created on the fly.
Batch Processing
Select any number of items and classify them all in one operation. Processing is sequential to respect API rate limits.
Context Menu Integration
A new "AI 分类论文" menu item appears in the right-click context menu for selected library items.
Preferences Pane in Zotero Settings
Plugin settings are registered directly in the native Zotero Edit → Preferences dialog, including:
| Setting | Default | Description |
|---|---|---|
| API Key | (empty) | Your DeepSeek API key |
| Model | deepseek-chat |
AI model to use for classification |
| Collection Root | AI主题分类 |
Root collection name in your library |
| Keep Original Collections | Off | If enabled, papers are added to the new collection without being removed from their current ones |
Built-in API Key Validation
A "Verify API Key" button in the settings pane sends a lightweight test request to the DeepSeek endpoint and displays the result inline (success / error message).
DeepSeek Reasoner Fallback
When the deepseek-reasoner model returns an empty response body (a known edge case for short outputs), the plugin automatically retries with deepseek-chat to ensure classification always completes.
Multi-Size Plugin Icons
The plugin ships with icons at 16 × 16, 32 × 32, 48 × 48, 96 × 96, and 128 × 128 pixels. The icon is displayed in the Zotero preferences sidebar alongside the plugin name.
Installation
Standard (Zotero 8 stable)
Install paper-classifier-1.0.0.xpi via Tools → Add-ons → ⚙ → Install Add-on From File…
Beta / Nightly Zotero
Install paper-classifier-1.0.0-beta-compatible.xpi instead.
This build uses a relaxed version range (strict_min_version: 6.999, strict_max_version: 9.9.*) to bypass the compatibility warning shown by Zotero beta builds.
Known Limitations
- Items without a title or abstract are skipped and reported in the failure summary. The DeepSeek API requires at least a title to generate a classification.
- The plugin uses the DeepSeek Chat Completions API and requires an active internet connection and a valid API key from platform.deepseek.com.
- The preferences pane registers asynchronously after Zotero starts. If the pane does not appear immediately, wait a few seconds and reopen the Preferences dialog.
- The
deepseek-reasonermodel may incur higher token usage. For most classification tasks,deepseek-chatis sufficient and more cost-effective.
File Checksums
| File | Description |
|---|---|
paper-classifier-1.0.0.xpi |
Standard release for Zotero 8 stable |
paper-classifier-1.0.0-beta-compatible.xpi |
Relaxed version range for Zotero beta builds |
Changelog
| Version | Date | Summary |
|---|---|---|
| 1.0.0 | 2026-02-24 | Initial public release |
Built with the Zotero 8 Bootstrap extension model and the DeepSeek Chat Completions API.