Turning Reviews into Real Insights: Emotion-Aware Sentiment Analysis with GraphRAG and Neo4j

Session Track: Data Intelligence

Session Time: 08:00 am - 08:15 am November 6

Session description

Passenger reviews are packed with emotional nuance, but most sentiment tools miss the big picture. In this lightning talk, Lijimol George presents GraphRAG, a hybrid framework powered by Neo4j that transforms unstructured airline reviews into structured, emotion-aware knowledge. The approach models service-specific aspects, such as seat comfort, cabin staff service, and in-flight entertainment, and incorporates emotional signals based on Ekman’s model of emotions, enabling fine-grained sentiment interpretation. By combining Sentence-BERT for semantic similarity, Cypher for graph querying, and RAG with LLMs for summarization, GraphRAG bridges the gap between raw narrative feedback and actionable, aspect-level insights. The talk illustrates how developers, data analysts, and graph enthusiasts can build similar pipelines by integrating emotional intelligence and aspect-level analysis into Neo4j, leveraging graph embeddings for hybrid search, and orchestrating structured retrieval with generative models. This domain-agnostic methodology can be extended to a wide range of sectors, including healthcare, public services, and e-commerce. By the end of the session, you’ll walk away with a clear understanding of how to set up a GraphRAG stack, query it effectively, and convert free-form text into emotionally intelligent, aspect-level insights—ready for real-world applications.