TaxoConf
A taxonomy-driven framework for conference organization. TaxoConf constructs a hierarchical taxonomy over accepted papers using LLMs, then leverages this structure to support four key functions.
01
Paper–reviewer assignment
Represents reviewer expertise as distributions over taxonomy nodes derived from publication history. Matches submitted papers to reviewers with high topical alignment.
02
PC member discovery
Identifies promising additional reviewers from a large-scale expert pool when the existing PC lacks sufficient coverage for specific topics or papers.
03
Oral session organization
Groups papers into coherent oral sessions by exploiting the taxonomy hierarchy, jointly optimizing capacity constraints and inter-session diversity.
04
Poster session organization
Arranges posters by combining taxonomy-based topical proximity with spatial layout optimization, placing related posters in adjacent positions.
Input Data
TaxoConf accepts paper metadata in both CSV and JSON formats. The required fields are id, title, and authors. An optional abstract field improves topic classification.
id,title,authors,abstract
1,"Retrieval-Augmented Generation with Adaptive Passage Selection","Alice Wang, Bob Chen, Carol Liu","We propose a novel RAG framework that dynamically selects..."
2,"Fair Exposure in Conversational Recommendation Systems","David Kim, Eva Martinez","This paper addresses fairness constraints in multi-turn..."
3,"Efficient Dense Retrieval via Learned Index Structures","Frank Zhou, Grace Tan","We introduce a learned indexing approach for approximate..."
4,"Query Performance Prediction Using Large Language Models","Hao Li, Isabel Fernandez","We study the effectiveness of LLMs as zero-shot query..."
5,"User Simulation for Evaluating Interactive IR Systems","James Park, Karen Wu, Leo Zhang","We present a user simulation framework grounded in..."
Output data
TaxoConf produces results in multiple formats to suit different workflows.
CSV
Session lists and schedules
HTML
Interactive pages for browsing
Excel
Spreadsheets using conference template
How to use
1. Select Oral Session Organization or Poster Session Organization in the sidebar.
2. Configure session parameters (tracks, time slots, capacity) in the setup panel.
3. Click Run to generate the schedule with LLM-powered session naming.
4. Edit session metadata (chairs, times, rooms) inline or via the detail panel.
5. Save Progress to preserve edits, Load Progress to resume later.
6. Export to Excel, CSV, or HTML when ready.
2. Configure session parameters (tracks, time slots, capacity) in the setup panel.
3. Click Run to generate the schedule with LLM-powered session naming.
4. Edit session metadata (chairs, times, rooms) inline or via the detail panel.
5. Save Progress to preserve edits, Load Progress to resume later.
6. Export to Excel, CSV, or HTML when ready.
Project page and materials
Contact
We welcome any feedback, feature requests, or collaboration inquiries. Please reach out via
GitHub Issues or email us at
daominji@student.rmit.edu.au.
Setup Panel: Paper Assignment
Mode
Assignment Mode
Loading assignment data summary from the server...
Data
Conference Demo Data
Demonstration mode uses the papers, reviewers, meta-reviewers, affinity matrices, and conflict matrices stored on the server.
Upload your own JSON data to run assignment in normal mode.
Parameters
Reviewer Workload
Reviewer Coverage
Meta-Reviewer Workload
Main Panel: Assignment Result
Paper View List
Assignment Detail
Run assignment and click a paper or PC member to inspect details.
Setup Panel: Reviewer Discovery
Status
Reviewer Discovery is still in progress. The backend integration for candidate retrieval and gap
analysis is being revised and the interactive controls are intentionally disabled for this build.
Planned Scope
The final version will surface reviewer-demand gaps, search candidate reviewers from external
sources, and let users inspect how recommended additions reduce uncovered topical demand.
Main Panel: Reviewer Discovery
Still In Progress
This tab is intentionally disabled in the current demo build. No search or recommendation results are
available yet.
Loading server-side presentation data...
Session Parameters
Parallel Sessions (M)
Time Slots (N)
Max per Session
Min per Session
Capacity Check
Checking capacity against the demo paper set...
Export
Export Format
Click a session cell after the run to inspect and edit.
Oral Schedule Result
Loading server-side presentation data...
Floor Plan Layout
Floor Plan Category
Specific Floor Plan
Board Count
Rectangle Rows
Rectangle Columns
Session Parameters
Session Count
Checking capacity...
Export
Export Format
Click any poster session card or board to inspect and edit.
Main Panel: Poster Arrangement Result
Token & Cost Statistics
Track LLM token usage and estimated costs across runs, workspaces, and globally.
Last Run
No run data yet.
Current Workspace
No workspace data.
Global (All Workspaces)
No global data.
Run History
No runs recorded yet.
Settings
Configure LLM provider, organization methods, and similarity engine. Changes take effect immediately.
LLM Provider & Model
Environment variables by provider:
Google Gemini:
Anthropic:
Google Gemini:
GOOGLE_API_KEY |
OpenAI: OPENAI_API_KEYAnthropic:
ANTHROPIC_API_KEY |
xAI: XAI_API_KEY