Nichirin: A custom search framework featuring augmented generation with retrieval capabilities.

Overview

Nichirin serves as an advanced layer atop Apache Solr, facilitating seamless data indexing operations.

1. What is Nichirin?

   • Nichirin acts as a surface or layer on top of Apache Solr, making data indexing a breeze.
   • It abstracts away the complexities of Solr indexing, allowing users to focus on providing their data without worrying about the nitty-gritty details.

2. Key Features:

   • Multi-level Crawling: Performs multi-level web crawling utilizing a depth-first search methodology, with text indexing and retrieval facilitated through Apache Solr.
   • Efficient Indexing: Integrated Apache Spark for parallel processing of URLs, improving the scalability and efficiency of both web crawling and text indexing.
   • Python package: Available as a Python package on PyPI for easy installation and integration

Commands

• install-solr to install solr
• create-core --core <core name> to create solr core,
• partition-data --path <path to the dataset> to partition the data
• pipeline --path <path to the dataset> generate embeddings of the partition data
• index-solr --data-path <path to dataset> --core <core to which the data needs to be sent> index the data
• query-solr --input_sen <input sen> --core_name <core name to query from> query the data from solr
• seed-urls --core <core name> --urls <urls separted with commas> to add the seed urls
• start-crawler to start the web crawler
• start-serve to start the web server

Quickstart

1. Begin by executing the install-solr command to install the Solr application.
2. Next, create the cores using the create-core command.
3. After setting up Solr and creating the cores, add seed URLs by running the seed-urls command.
4. Once the seed URLs are added, initiate the crawling process with the start-crawler command. Be patient, as this step may take some time.
5. Finally, to view the results, launch the Flask web app using the start-serve command.

Contributing and Feedback: We welcome contributions! If you’d like to enhance Nichirin or report issues, feel free to submit a pull request. For feedback or questions, open an issue on our GitHub repository.