Tools exist online to make literature review a less daunting endeavour. With the increased adoption of AI, more and more are powered with AI. What if you could further digest selected papers based on dynamic queries from your command-line?
paper-selector is an AI-powered CLI tool to filter and prioritize pdf documents, especially research articles based on natural language queries. It uses hybrid search (semantic + keyword search) to help you further digest your papers or documents.
paper-selector is written in Julia. You must first install Julia on your system. You can refer to Julia documentation to get started.
A number of additional packages is also required to run the tool. Install them by running the following command.
julia setup.jlTo use paper-selector, you need to:
- Prepare your documents or papers in an input directory
Add all the documents you want to digest in the same directory. We only support PDF for now.
- Specify parameters if needed
You can specify the path to your input directory and other parameters in your Unix environment with:
export INPUT_PATH=<actual-path>
export OUTPUT_PATH=<actual-path>
export NUM_QUERIES=<actual-value>
export TOP_RESULTS=<actual-value>
export MODEL_NAME=<huggingface-model-name>If those parameters are not defined, paper-selector looks by default for documents in a directory named input in the project home directory. Make sure that directory exists and contains your papers. Other parameters are set by default. You don't need to provide them unless you want to have more control.
- Run the tool
Once you have everything set up, you can run the tool using the commands below:
cd src
julia select.jl- Enter your queries
Next, enter your queries when prompted. The default number is 5 but you can change this with the corresponding parameter above. For best performance, we recommend passing in queries that sound almost like English sentences. Once done, wait for the tool to run.
- Check the results
Check the results in the directory output from the project home directory :)
We are working hard to add more features!