Search Engine for Atlassian Tools
Project Goal:
Develop an AI-powered search engine to enhance Jira and Confluence’s search functionality. The system will provide precise results, summaries of pages and tickets, and an improved user experience by leveraging AI and semantic search capabilities.
Tools and Technologies
- Search Indexing and Retrieval:
- Elasticsearch: Open-source search engine for full-text search and indexing.
- FAISS: Performs semantic similarity searches using embeddings.
- AI Models:
- Summarization & Q&A: HuggingFace models like Flan-T5 or BART.
- Sentence Embedding: Models like sentence-transformers for semantic matching.
- Tokenizer Model: BERT tokenizer to extract and index key topics.
Integration
- Data Sources: Jira and Confluence APIs for fetching issues and content.
- Backend: Python (Flask) or Java (Spring Boot) for data processing and retrieval.
- Frontend:
- Web Portal: React.js or Angular for search interaction.
- Browser Extension: Direct search integration in Jira/Confluence UI.
- Jira/Confluence Plugin: Built using Atlassian Forge or Connect frameworks.
Retrieval-Augmented Generation (RAG)
Combines document retrieval with generative models to answer queries in real time:
- Fetch updates using Jira/Confluence APIs.
- Retrieve relevant documents via Elasticsearch/FAISS.
- Augment with generative models (e.g., GPT or Flan-T5) for real-time, context-aware answers.
Workflow
Example Workflow:
Query → Retrieve Top N Relevant Documents → Augment with Generative Model → Provide Real-Time Answers.