Graph Data – Gemini Data

Graph Databases: No Code vs Low Code vs Code

Jenn Snider — Wed, 09 Aug 2023 12:07:47 +0000

Gemini Data’s resident data scientist and bioinformatician Sixing Huang recreates the same graph data use case in various architectures with varying levels of technical skill for end users.

"Knowledge graphs are gaining traction quickly. They store information like a human, that is, via subject-verb-object triples. We can transform many existing data into knowledge graphs and then learn a lot by exploring, searching, and analyzing them."

Huang then walks through the major steps in preparing the data to be transformed into a series of subject-verb-object triples in CSV that gets imported into Neo4j. He also imports this data into both Neo4j Desktop and AuraDB and compares and contrasts ease of use and computing power.

Read Sixing Huang’s full tutorial on Medium.

Creating Context-Aware Chatbots with ChatGPT, Knowledge Graphs, Neo4j, and Gemini Explore

Jenn Snider — Thu, 27 Jul 2023 12:31:30 +0000

Gemini Data’s resident data scientist and bioinformatician Sixing Huang dives into creating a dictionary chatbot using ChatGPT, Neo4j, and Gemini Explore:

"In the Japanese language, onomatopoeic words, including “giongo (擬音語)”, “giseigo (擬声語)” and “gitaigo (擬態語),” are unique expressions that vividly depict sounds, actions, and feelings. These words are abundant in Japanese culture and are used in various contexts, including literature, manga, anime, and everyday conversations. But onomatopoeic words are hard for foreigners to learn. You cannot deduce their meanings from the spellings most of the time. For example, the word コツコツ (kotsukotsu) means “laboriously, steadily”, while its look-alike ゴツゴツ (gotsugotsu) means “gnarled, rugged”. And the word ゴホゴホ (gohogoho) represents hacking cough, even though its pronunciation does not sound like coughing at all. It takes time, examples, and lots of practice to internalize even the basic ones. And there are 1,190 of them in the JapanDict."

Huang then outlines how to create a chatbot to help him master the onomatopoeic words in the Japanese language using Neo4j, AuraDB, Gemini Explore, and OpenAI.

Read Sixing Huang’s full tutorial on Medium.

Clinical Trials as Graphs and Vectors

Jenn Snider — Mon, 24 Jul 2023 14:23:59 +0000

Gemini Data’s resident data scientist and bioinformatician Sixing Huang creates a search engine of clinical trials data with Neo4j, Gemini Explore, and Qdrant.

“As a demonstration, 24 Japanese clinical studies were downloaded from the site. In addition, the English SNOMED CT provides me with the taxonomies for medical conditions and body structures. On the one hand, I used Apache Hop to import the data into the graph database Neo4j. On the other hand, the trial descriptions were embedded and then inserted into the vector database Qdrant. Users can not only search trials semantically on Qdrant, but also learn their details, relationships, and statistics on Neo4j and Gemini Explore. This two-database setup allows users to quickly answer the three example questions above.”

By improving the searchability of clinical trials data providers can ask complex questions of their data sets like, “What trials are targeting leukemia?” and “What trials compare the drugs iberdomide and lenalidomid?”

Read the full walkthrough on Medium.

Unlocking New Horizons: 4 Perspectives on Leveraging LLMs and Graph Databases

Jenn Snider — Wed, 12 Jul 2023 20:49:10 +0000

In the vast landscape of information technology, Large Language Models (LLMs) and graph databases have emerged as powerful tools revolutionizing how we process and analyze data. LLMs, such as OpenAI’s GPT-X systems powering the wildly popular ChatGPT, have made significant strides in natural language understanding and generation, while graph databases offer a flexible and efficient way to represent and query complex relationships. When these two technologies are combined, they unlock unprecedented possibilities for knowledge extraction and decision-making. Let’s explore four ways of thinking about LLMs and graph databases, shedding light on their potential applications and synergies.

Natural Language Understanding and Generation

LLMs are designed to comprehend and generate human-like text, making them invaluable for natural language understanding (NLU) and natural language generation (NLG) tasks. By leveraging LLMs in conjunction with graph databases, we can enhance the capabilities of traditional query systems. Graph databases, like Gemini Explore, provide a rich representation of connected data with nodes and edges, capturing relationships and context. This enables LLMs to generate more accurate and context-aware responses by taking into account graph-based context.

For example, imagine a customer support chatbot that utilizes a graph database to store information about customer profiles, products, and common support issues. By integrating an LLM, the chatbot can understand and generate responses in a more conversational manner, drawing insights from the graph structure to provide personalized and contextually relevant solutions.

Knowledge Graph Enrichment

Graph databases excel at representing and connecting heterogeneous data sources. LLMs can be employed to enrich knowledge graphs by extracting structured information from unstructured data. We can create powerful knowledge extraction pipelines by training LLMs on domain-specific corpora and integrating them with graph databases.

Consider a healthcare application that stores patient records in a graph database. By applying LLMs to unstructured clinical notes, the application can extract structured information such as diagnoses, medications, and treatment plans. This enriched knowledge graph can then be leveraged for advanced analytics, medical research, and personalized patient care.

Recommendation Systems

Graph databases provide a natural framework for modeling and querying complex relationships, making them an excellent choice for building recommendation systems. With their ability to understand user preferences and generate relevant suggestions, LLMs can enhance the accuracy and personalization of these recommendation systems.

By combining the power of LLMs and graph databases, we can create recommendation engines that consider both explicit and implicit user preferences. For instance, a movie streaming platform can leverage a graph database to model user interactions, such as watched movies, ratings, and social connections. By utilizing an LLM, the platform can generate personalized movie recommendations based on the user’s viewing history, preferences of similar users, and other relevant contextual information from the graph.

Complex Network Analysis

Graph databases serve as a foundation for complex network analysis, enabling us to uncover patterns and insights from interconnected data. When coupled with LLMs, this analysis becomes even more powerful as LLMs can identify complex patterns in large-scale networks.

For example, social media platforms can employ LLMs and graph databases to detect and understand misinformation spread across their networks. LLMs can identify potentially misleading information by analyzing the text content of posts, comments, and shared articles. The graph structure of the social network can be used to track the propagation of such content and identify influential nodes. This integrated approach helps platforms proactively combat misinformation and protect their user base.

Conclusion

The combination of LLMs and graph databases presents a multitude of opportunities across various domains. The potential applications are vast, from improving natural language understanding and generation to enriching knowledge graphs, enhancing recommendation systems, and enabling complex network analysis. By leveraging the strengths of LLMs and graph databases, organizations can unlock new horizons in data analysis, decision-making, and user engagement.

The synergy between LLMs and graph databases enables a deeper understanding of complex relationships and context. It empowers systems to provide more accurate, personalized, and context-aware responses. Whether it’s in customer support, healthcare, recommendation systems, or combating misinformation, integrating LLMs and graph databases offers a paradigm shift in data processing and analysis.

By embracing these technologies and exploring the four perspectives mentioned above – natural language understanding and generation, knowledge graph enrichment, recommendation systems, and complex network analysis – organizations can unlock new insights, improve decision-making, and deliver enhanced user experiences in today’s data-driven world. The journey toward harnessing the full power of LLMs and graph databases has just begun, and it’s an exciting path to be on.

Ready to get started? Gemini Explore integrates with the latest LLM, GPT, and machine learning advancements for enterprises aiming to leverage the potential of generative AI. We offer a quick, secure, and effective way to integrate LLM technology with your enterprise data to deliver actionable insights and recommendations to drive improved business outcomes.

Book Review: Knowledge Graphs

Jenn Snider — Thu, 29 Jun 2023 11:58:28 +0000

Gemini Data’s resident data scientist and bioinformatician Sixing Huang dives into the new book by Jesús Barrasa, Amy E. Hodler, and Jim Webber, Knowledge Graphs: Data in Context for Responsive Businesses:

"Within merely 79 pages, the authors have covered the key aspects of KG, including the definition, construction, types, and its roles in contextual AI and business digital twins. Both technical and non-technical readers can enjoy the book because it focuses on the core concepts and spares us the programming details. In this article, I would like to share some of my learnings."

Knowledge Graphs has three main sections:

Taxonomy and ontology. The book addresses the potential of enhancing property graph models by adding taxonomy, which helps organize nodes in a broader-narrower hierarchy.
Taking action and making decisions. By classifying knowledge graphs into two types – “actioning,” which is used for data management, and “decisioning” which are used for analytics and data science.
Digital twin. The authors outline the role of knowledge graphs in creating digital twins, virtual models of real-world problems that can be used for analysis and forecasting.

The authors stress the importance of practical experience in mastering knowledge graphs and mention the availability of many practical articles on Medium to guide users in building and evaluating knowledge graphs.

Read Sixing Huang’s full review of Knowledge Graphs on Medium.