ATOM

• Complete Architectural Redesign: ATOM now employs a three-module parallel pipeline for DTKG construction and updates.
• Atomic Fact Decomposition: A new first module splits text into minimal "atomic facts," addressing the "forgetting effect" where LLMs omit facts in longer contexts.
• Enhanced Exhaustivity and Stability: The new architecture achieves significant gains: ~31% in factual exhaustivity, ~18% in temporal exhaustivity, and ~17% in stability.
• Dual-Time Modeling: Implemented dual-time modeling (t_obs vs. t_start/t_end) to prevent temporal misattribution in dynamic KGs.
• Parallel 5-Tuple Extraction: Module-2 now directly extracts 5-tuples (subject, predicate, object, t_start, t_end) in parallel from atomic facts.
• Parallel Atomic Merge Architecture: Module-3 uses an efficient, parallel pairwise merge algorithm, achieving 93.8% latency reduction vs. Graphiti and 95.3% vs. iText2KG.
• LLM-Independent Resolution: Replaced slow LLM-based resolution with distance metrics (cosine similarity) for scalable, parallel merging.

iText2KG v0.0.9

Improvements

• We correct the bug #38 in neo4j storage.

iText2KG v0.0.8

New Features

• iText2KG_Star
  • Direct relationship extraction (faster)
  • Eliminates separate entity extraction step
  • No isolated/invented entity handling
• Dynamic Knowledge Graphs
  • Dynamic KGs with temporal tracking
  • Incremental updates with existing graphs
• Facts-Based Construction
  • Structured fact extraction via Document Distiller
  • More exhaustive knowledge graphs

Improvements

• Async Migration: All methods now async/await
• Enhanced Logging: Structured logging system
• Universal LangChain Support: All chat/embedding models
• Better Error Handling: Production-ready reliability

Technical

• Enhanced relationship extraction
• Python 3.10+ compatibility
• Comprehensive examples

We corrected the following issues: #34, #33, #29, #28, #26, #22

Refactoring the iText2KG code

-The entire iText2KG code has been refactored by adding data models that describe an Entity, a Relationship, and a Knowledge Graph.

• Each entity is embedded using both its name and label to avoid merging concepts with similar names but different labels, such as Python: Language and Python: Snake.
• The weights for entity name embedding and entity label are configurable, with defaults set to 0.4 for the entity label and 0.6 for the entity name.
• A max_tries parameter has been added to the iText2KG.build_graph function for entity and relation extraction to prevent hallucinatory effects in structuring the output. A max_tries_isolated_entities parameter has been added to the same method to handle hallucinatory effects when processing isolated entities.

Latest update

• Fixing the bug reported in #7.
• Update the iText2KG build_graph function to perform matching between newly constructed graphs and existing graphs after the construction process.

Supporting other LLMs

• Now, iText2KG is compatible with all the chat/embeddings models LangChain supports. (#1)
  The constructed graph can be expanded by passing the already extracted entities and relationships as arguments to the build_graph function in iText2KG.
• iText2KG is compatible with all Python versions above 3.9. (#2)
• Some bugs in the overall architecture are corrected.

Latest update

• Updating the version of neo4j in the requirements.
• Adding datasets for threshold estimation.
• Adding the paper link.

Fixing the workflow

V0.0.2

Update dependencies in publish_on_pypi.yml and setup.cfg

First Release

V0.0.1

Add PyPI publishing workflow and configuration files