Skip to main content

Tags and classification

Chive uses a combination of hierarchical fields and user-contributed tags to organize eprints. This guide explains how to tag papers and navigate the classification system.

Fields vs tags​

Chive has two classification systems:

SystemWho controls itPurpose
FieldsCommunity-governed taxonomyPrimary hierarchical classification
TagsIndividual usersFlexible, emergent folksonomy

Fields​

Fields form a hierarchical taxonomy similar to arXiv categories:

Authors assign fields when submitting. Community proposals can add new fields or reorganize the hierarchy.

Tags​

Tags are free-form labels anyone can add:

  • attention-mechanisms
  • transformer-architecture
  • benchmark-dataset
  • reproducibility-study

Tags help with discovery beyond the formal taxonomy.

Adding tags​

To your own papers​

  1. Go to your eprint's page
  2. Click Add Tags
  3. Type a tag name
  4. Select from suggestions or create a new tag
  5. Click Save

To any paper​

If you've read a paper and want to help classify it:

  1. Click Add Tags on the paper's page
  2. Your contributed tags appear with your name
  3. Popular tags get promoted; rarely-used tags fade

Tag guidelines​

Good tags​

  • Specific methodologies: monte-carlo-simulation, transformer-xl
  • Datasets: imagenet, squad-2.0
  • Applications: drug-discovery, climate-modeling
  • Paper types: survey-paper, negative-results, reproducibility

Avoid​

  • Too broad: science, research, paper
  • Duplicating fields: Don't tag machine-learning if the field is already cs.AI.ML
  • Personal notes: to-read, interesting, read-later
  • Promotional: groundbreaking, must-read

Tag moderation​

Spam detection​

Chive automatically filters:

  • Promotional or marketing tags
  • Offensive content
  • Tags unrelated to the paper's content
  • Bulk tagging by bots

Community curation​

  • Tags with many contributors rise in visibility
  • Rarely-used tags are de-emphasized
  • Trusted editors can remove inappropriate tags
  • Disputed tags can be flagged for review

PMEST classification​

Behind the scenes, Chive maps papers to the PMEST framework:

DimensionWhat it capturesExamples
PersonalityWhoAuthors, institutions, funders
MatterWhatSubject field, methodology
EnergyHowResearch type (theoretical, empirical, review)
SpaceWhereGeographic focus, language
TimeWhenPublication date, historical period studied

PMEST enables faceted search: "Show me empirical AI papers from European institutions published in 2024."

Proposing new fields​

If the taxonomy lacks a field for your research area:

  1. Go to Governance → Proposals
  2. Click Propose New Node
  3. Select field as the node subkind
  4. Fill out the form:
    • Proposed field ID (e.g., cs.QML)
    • Label (e.g., "Quantum Machine Learning")
    • Alternate labels (synonyms like "QML")
    • Description and scope
    • Justification
  5. Optionally propose edge relationships to parent fields
  6. Submit for community voting

Voting requirements​

Proposal typeApproval thresholdMinimum votes
New field67%5 votes, 3 from field experts
Update field60%3 votes
Merge fields67%5 votes

View which tags are gaining popularity:

  1. Go to Trending in the navigation
  2. Select a time window: 24 hours, 7 days, or 30 days
  3. See tags ranked by recent usage growth

Tags are scored based on:

  • Recent usage count: Number of times the tag was applied in the time window
  • Quality score: Considers uniqueness and relevance
  • Growth rate: Comparison to previous period

API access​

GET /xrpc/pub.chive.tag.getTrending?timeWindow=7d&limit=20

Time windows: 24h, 7d, 30d

Searching by classification​

Field browsing​

  1. Go to Browse → Fields
  2. Navigate the hierarchy
  3. Click a field to see all papers
  1. Click any tag on a paper
  2. Or search: tag:transformer-architecture
  3. Combine with other filters: tag:reproducibility field:cs.AI

Faceted filtering​

Use the sidebar filters:

  • Field: Select from hierarchy
  • Tags: Multi-select popular tags
  • Research Type: Theoretical, empirical, review, etc.
  • Date Range: Filter by publication date