RelateTime

English · 简体中文

RelateTime is an Obsidian plugin designed for creators, web novel writers, and worldbuilders. Built with React and a high-performance D3.js physics engine, it turns dry Markdown text properties into a dynamic character relationship graph that evolves over time.

This plugin was created by an Agent.

With its original “hybrid deterministic engine + LLM-based global knowledge graph completion,” RelateTime can automatically complete complex kinship, fellow-disciple, and in-law relationships that you have not written out, and warn you when logical or gender contradictions occur.

✨ Features

🕰️ Full-Scale Ultra-Precision Timeline

• BC/BCE support: Whether it is prehistoric times, a cultivation era, or an interstellar future, it supports an unlimited span from 3000 BC to the future.
• Seamless dynamic zoom: Smoothly zoom from “century view (100 years)” all the way down to “hour view (down to a specific date and hour).”
• Synchronized state transitions: Drag the timeline and the graph displays the character relationships that are exactly active at that point in time.

🧠 Hybrid Inference Engine

• 100% stable local rule engine: Automatically infers mutual sibling relationships, shared-parent relationships, shared masters among fellow disciples, and other deterministic logic with zero latency and absolutely consistent results.
• LLM Integration for global graph completion: Connect to an OpenAI-compatible LLM API (such as gpt-4o or a local LM Studio instance). The model analyzes the entire character network and automatically infers cross-generational (grandparent–grandchild / master's master), collateral (uncles/aunts), and in-law (elder brother's wife / parents-in-law / wife's elder brother) relationships.
• Two inference modes: Choose “Show partial inferences (core direct-line and closed-loop relationships only)” or “Show all inferences (including collateral and cross-generational relationships)” to avoid an explosion of edges.

🛡️ Intelligent Conflict Detection

• Automatically verifies whether the “gender” property in a note matches the recorded relationship label.
• When a logical contradiction occurs (for example, gender is set to male but the character is recorded in the wife role; or generation/seniority relationships conflict and overlap), the graph highlights the edge with a red warning line. Hover over it to view a detailed conflict analysis tooltip.

🎨 Optimized D3 Geometric Rendering

• Flow-slot anti-overlap layout: Multiple relationship edges (such as two people being both spouses and master-disciple) are automatically laid out in parallel at fixed intervals, and labels are staggered along the edge direction to completely eliminate text overlap.
• Smart single-line bidirectional merging: Reciprocal identical relationships (such as “friend ⇄ friend” and “fellow disciple ⇄ fellow disciple”) are intelligently collapsed into a single solid line with bidirectional arrows, keeping the graph clean and clear.
• Deeply customizable appearance: Fully restores and enhances Obsidian's native graph physics settings (centering force, repulsion force, link distance, etc.), and supports pixel-level adjustments for node size, line width, and opacity at each hierarchy level.

📦 Installation

RelateTime is now available in the Obsidian Community Plugin marketplace. You can search for RelateTime and install it directly in the app.

Manual installation:

1. Go to the Releases page of the GitHub repository.
2. Download the latest RelateTime.zip.
3. Extract it and place the relatetime folder into the .obsidian/plugins/ folder in your Obsidian vault.
4. Restart Obsidian and enable RelateTime under Settings -> Third-party plugins.

🚀 Usage

1. Complete character cards (Frontmatter)

In your Obsidian notes, make sure to include the following YAML Frontmatter properties:

---
type: 人物
character-name: 温迪
性别: 男
关系: 张三 - 师父（2020-08-13 BC 1:00 ～ 2024-08-26 BC 17:00），李四-夫妻（2025-08-13 BC 13:00:00 ～ 2025-08-26 BC 13:00:00）、赵五 -大哥（固定）
---

(Note: Both the property name and property value of type can be customized in the plugin settings. The example above uses the default Chinese property names.)

Relationships:

Dates may be CE/AD (without BC), may omit hours and minutes, and relationships that do not change over time can be marked with (固定) (fixed).

For example, 张三 - 师父（2020-08-13 BC 1:00 ～ 2024-08-26 BC 17:00） means that from 2020-08-13 BC 1:00 to 2024-08-26 BC 17:00, 张三 is the “master” of this file.

2. Configure the LLM (for global deep inference)

• Open the RelateTime plugin settings page.
• Find LLM Relationship Inference Settings.
• Enter your API URL (local model addresses are supported) and API Key.
• Enter the model name (such as gpt-4o-mini), then click Hi Test (hi测试) to verify connectivity.
• Click AI Re-infer Relationships (AI 重新推演关系) in the upper-left corner of the graph to generate high-level global relationships in one click.

3. Operate the graph and list

• Click the Relationship Network icon in the left sidebar to open the view.
• Drag the bottom ruler to pan through time.
• Scroll the mouse wheel to zoom in/out on the timeline precision.
• Click a node to lock focus. The right-side menu provides powerful search (showing neighbors) and appearance customization.
• List view: In the list page, a relationship entry indicates that the character at the top is the [relationship] of the character on the left.