GraphGrid Changelog

[2024-03-01]

2.0.0 Major Release

Version 2.0.0 of GraphGrid includes the following major fixes and improvements. GraphGrid recommends that all instances of 1.4.x be updated with this version immediately.

Manager

• Manager Model UI enables users to create and maintain their graph data model
• New frontend capabilities to create and update Node types, Edge types, and Properties
• Associate Edge types and Properties with Nodes; Associate Properties with Edge types
• Adds Showme (user created queries) support in UI
• Improves UX for graph visualization, navigation and operations performed to the Manager Model

NLP

• Batch Inferencing on NLP Data Extraction for improved performance
• GPU Support for NLP Data Extraction for improved performance
• Introduces NLP HelperModel
  • Communication between nlp-model-training (python) and NLP Module (java)
  • More consistent use of tensorflow capabilities
• NLP Model Training & Airflow
  • Train your own NLP models
  • Use DAGs to control training workflow
  • Bring your own data to train and finetune NLP Models
  • Create new DAGS for ETL pipelines to ingest data
  • Multi-GPU Training Support
  • Auto-detection of GPU Memory and optimization for running concurrent jobs
• NLP Model Training SDK
  • Programmatically interact (python) with GraphGrid NLP Model Training
  • One-file format supports training datasets of different NLP tasks simultaneously
• Improves similarity clustering algorithm (clusters similar text documents)
• Improves Binary and Categorical Sentiment Analysis models
• Introduces Sentence Boundary Detection ML Model (improved from rules-based sentence splitting)

Packaging

• Adds module enable/disable commands
• Adds GPU commands (gpu check; gpu enable/disable)
• Improves secure password generation
• Get/Set GraphGrid Config
• Linux support and security improvements
• QoL Improvements
• Adds GraphGrid user/group 9191 supporting consistent permissions management

Docs

• Switch over to Docusaurus
• Adds Airflow Tutorial
• Adds NLP Model Training SDK Tutorial
• Adds Machine Learning Basics and Batch Inference Tutorial

General

• Improves module migrator support

[2022-09-01]

1.4.6 Patch Release

Version 1.4.6 of GraphGrid includes the following fixes and improvements. GraphGrid recommends that all instances of 1.4.x be updated with this version immediately.

• Adds delete endpoint for notifications in Publish
• Adds invoiceTransaction details to the getInvoices endpoint in Payment
• Fixes bug regarding case mismatch between mention objects and mentionIds which resulted in relationships rarely ever making it into the graph
• Removes inclusion of erroneous NO_RELATION (or ELATION as it was seen in the graph) relationship between mentions within the graph

[2022-02-16]

1.4.5 Patch Release

Version 1.4.5 of GraphGrid includes the following fixes and improvements. GraphGrid recommends that all instances of 1.4.x be updated with this version immediately.

• Updates log4j to 2.17.1 to further mitigate Log4jShell vulnerabilities
• Fixes issue where graphgrid startup fails when the data directory includes spaces
• Adds more detailed line items for PayPal transactions

[2021-12-16]

1.4.4 Patch Release

Version 1.4.4 of GraphGrid is an emergency patch release for the concern listed below. GraphGrid strongly recommends that all instances of 1 4.x be updated with this version immediately.

• Mitigated exposure to Log4j JNDI attacks targeting 3rd party components used in GraphGrid. All Log4j dependencies now use 2.16.1.

[2021-11-29]

1.4.3 Patch Release

Version 1.4.3 of GraphGrid includes the following fixes and improvements. GraphGrid recommends that all instances of 1.4.x be updated with this version immediately.

• Added restart on failure for all GraphGrid services
• Added RabbitMQ data volume mount to the GraphGrid data directory <data-dir>/rabbitmq/data
• Fuze: updated Replication test policies to use sendingUri instead of neo4jUri

[2021-10-15]

1.4.2 Patch Release

Required update for existing GraphGrid 1.4 instances.

Version 1.4.2 of GraphGrid includes the following fixes and improvements. GraphGrid recommends that all instances of 1.4.0/1.4.1 be updated with this version immediately.

• Search: fixed issue with search all expecting incorrect params
• NLP: improved TF calculation to not be limited to the last sentence for each mention
• Payment: per Authorize.net API specification we are now truncating transaction fields as follows:
  • 31 characters for itemID
  • 31 characters for name
  • 4 decimal places for unit price
  • 4 decimal places for quantity
• Install script now displays output messaging when restarted after an interrupted or incomplete install

[2021-09-28]

1.4.1 Patch Release

• Manager: resolved race condition during initial startup that may prevent Showmes from functioning properly
• NLP: addressed an issue with the cluster depth logic so that similarity clustering works as expected
• UI fixes/improvements:
  • Table view fixed and re-enabled
  • User menu initials center correctly
  • Save As model closes on submit and displays notification bar with status of success or failure
  • Improved UX for edit properties by adding action icons for save and cancel
  • Notifications no longer hide behind active panels
• Payment: added truncation for descriptions too long for Authorize.net
• Removed deprecated config properties:
  • /1.4/search/default/spring.aws.search.s3.bucketname
  • /1.4/search/default/spring.aws.search.s3.region

[2021-09-15]

1.4.0 Initial Release

Offered In Two Editions - Enterprise and Ecommerce

Manager

• Model Management
  • Create and manage a custom graph model that meets your existing and future data requirements with a library of node types, edge types, properties, and constraints to define the structure of the graph data model.
• Showme Management
  • Create and manage customized dynamic APIs that use Geequel to return a targeted result set of graph-based data. As new data is introduced, a Showme will incorporate that new data into results produced in real-time.
• Routes
  • Chain requests following the execution of a Showme.
• Request Sequencer
  • Chain Showmes and API endpoints in succession with data dependencies.

Search

• Text-Based Search
  • GraphGrid's Robust and efficient search capabilities leverage Elasticsearch index policies and include the ability to perform searches within the UI.
• Index Management
  • Creation and management of index policies to be used by Search.
• Elastic Scripting
  • Enable (expose via API endpoints) to use Elastic scripts within Search.

Payment Ecommerce Edition ONLY

• Create and manage customer payment profiles.
• Produce invoices that are based on Showme output.
• Includes payment processing via authorize.net and Paypal.
• Has the option for one-time "orders" without requiring a customer payment profile.
• Other tools to create a smart shopping cart experience.

Natural Language Processing (NLP) Enterprise edition ONLY

• Data Extraction
  • Data extraction (or Annotating text) is the process by which text is turned into a graph structure. Nodes representing text to be processed are used by GraphGrid's NLP to create a graphical representation of the original text. This representation inside the graph is the basis for all of NLP's other language processing features.

    - Data extraction using bidirectional encoder representations from transformers (BERT)
    - **Data extraction using CoreNLP Mode (to be removed in 1.5).**  
• Document Similarity Scoring
  • Similarity scoring is based on a term frequency-inverse document frequency (TF-IDF) method. Embed each document in a corpus based on their tf-idf vectors to enable similarity calculations between two documents. Term-frequency is calculated once per document and stored on the graph in HAS_TF relationships, while inverse-document frequency is calculated as needed.
• Document Summarization
  • A document's summary is composed of the k most relevant sentences. A sentence's relevancy is determined by the average of its word's tf-idf score (for normalization).
• Document Origination
  • Ability to take a document and form a story-arch: an ordered list of documents that come from one another.
• Paraphrase Detection
  • Two documents are embedded into sentence vectors and compared for paraphrasing of one another.
• Continuous Processing (CP)
  • Event-based processing of new nodes dropped directly onto the graph. Makes use of a message broker (RabbitMQ). Future plans include adding support for SQS and Kafka.
• Non-disruptive Interaction
  • NLP interacts with an existing graph without disrupting original graph structure
• Document Cards
  • Document cards provide summarization of an article to present details in UI.
• Multilingual Translation
  • Arabic
  • Chinese
  • French
  • German
  • Japanese
  • Korean
  • Portuguese
  • Russian
  • Spanish

Fuze

• Event Hub
  • Responds to changes in graph data and sends messages to a broker.
• Trigger Manager
  • Keeps triggers in sync and active across databases.
• Replication
  • Bi-directional movement of data between 2 instances to maintain consistency.
• Batch Data Writing
  • Offloads high volume data writing from the database to offer batch processing with error handling and retry capabilities.

Curator

• Graph Project Management
  • Projects are a means to organize and segment multiple analysis areas of focus and allows users to create/share/delete graph projects.
• Graph Database Management
  • Add/edit/delete nodes, properties, edges, and constraints in a graph database.
• Dashboard Navigation
  • Navigate the graph and utilize other UI capabilities like Search, Showmes, and NLP.