This Week in Neo4j: Embeddings, Algorithms, Docker, LangChain and more

GETTING STARTED WITH GRAPHS

• • GRAPHACADEMY: LLM Fundamentals
  • READ: ACID – Explaining Data Consistency
  • WATCH: Importing CSV Data with Neo4j
  • TRY: Neo4j AuraDB Free

  FEATURED NODES SPEAKER: Karina Isla-Rios

  
  

  Karina is a data passionate in the MedTech sector. She is interested in developing AI algorithms to improve the analysis of clinical data, disease management, clinical decision support and medical documentation.
  Connect with her on LinkedIn.

  In her session at NODES ” Graphs Enabling Glycoscience” together with Alexander Jarasch, Karina describes how graphs are used to represent clinically meaningful glycomics data and how Neo4j helps with managing complex biological data collections to help fight cancer.

  
  
   

  EMBEDDINGS: An Extremely Simple but Effective Way to Improve Search over Text Embeddings

  
  Fanghua presents a simple yet effective method to enhance similarity search in GenAI solutions like RAG, using Neo4j APOC ML procedures, and compares results across major cloud providers without relying on conventional techniques like model fine-tuning or prompt engineering. He includes examples and test results using embeddings from Azure OpenAI, Google VertexAI, and AWS Bedrock

   

  ALGORITHMS: Unlocking DAGs in Neo4j: From Basics to Critical Path Analysis

  At NODES, we introduced a few enhancements and additions to our Graph Data Science offering. New Directed Acyclic Graphs (DAGs) Algorithms were added: Longest Path and Topological Sort. Pierre Halftermeyer explores the essence of DAGs and their applications and demonstrates their potential through an in-depth analysis of the critical path of a Gantt chart.    

  
  

  DOCKER: Simple Graph Database Setup with Neo4j and Docker Compose

  Matthew Ghannoum wrote a tutorial that runs you through the steps necessary to get your Neo4j instance up: Creating the Docker Compose Files, Adding an Environment Variables File for your Password and Running your Neo4j using your Docker Compose file.

  LANGCHAIN: Hosted LangServe + LangChain Templates

  
  This video showcases a LangChain template from Neo4j: the neo4j-semantic-layer. LangChain shows us how to import the template and turn it into a web service with LangServe, then uses Hosted LangServe to deploy it and LangSmith for app observability. It introduces a general workflow that can be used with any LangServe template.

  
  

  TWEET OF THE WEEK: Tomaz Bratanic

  
  

  Do you want to finetune a text2cypher LLM but can’t find a dataset? Is there a new LLM you wish to evaluate for its Cypher generating abilities? The problem is that there are no publicly available text2cypher datasets that you could use. I want to change that.

  — Tomaz Bratanic (@tb_tomaz) January 25, 2024

  
  Don’t forget to share it if you like it!