If you use the code from this repository, please cite the following paper:
Arts S., Melluso N., Veugelers R. (2023). Beyond Citations: Text-Based Metrics for Assessing Novelty and its Impact in Scientific Publications. https://doi.org/10.48550/arXiv.2309.16437
This repository is dedicated to the assessement of novelty and its impact of scientific publications, employing Python scripts and Jupyter notebooks. It is designed with a dual purpose:
- Reproduce Results:
- Replicate the findings of the original paper, which analyzes data from the Microsoft Academic Graph (MAG), now OpenAlex, encompassing a comprehensive collection of papers from 1800 to 2020. The data can be accessed here: https://zenodo.org/record/8283353.
- Custom Analysis:
- Enable users to apply the analysis, including preprocessing and metrics calculation, to a tailored set of papers for individual research needs.
The methodology is systematically organized into the following segments:
- Data Collection
- Preprocessing
- Text Embeddings
- Cosine Distance
- New Word
- New Bigrams
- New Trigrams
- New Word Combinations
Each segment is integral for extracting text-based metrics to measure the novelty and its impact of scientific publications.
The repository contains scripts and detailed Jupyter notebooks that guide users through each step of the process. The notebooks are particularly beneficial for those aiming to execute specific tasks or a subset of the entire process.
0.tutorial: A comprehensive guide that offers a step-by-step walkthrough of all phases, serving as an introductory overview.1.data-collection: Instructions for downloading a custom set of papers from OpenAlex or searching within the Zenodo repository.2.preprocessing: A guide for preprocessing titles and abstracts (and full texts, if available) of a selected set of papers.3.text-embeddingsand4.cosine-distance: Notebooks for generating text embeddings and calculating cosine similarity.5.new-word,6.new-bigram,7.new-trigram,8.new-word-comb: Detailed guides for identifying new lexical elements and combinations in processed papers.
Users are encouraged to adapt the code for their specific research needs, ensuring a flexible and customizable approach to analyzing scientific novelty and impact. To this end the notebooks are organized as follow:
Contributions to enhance the code and extend its functionalities are warmly welcomed. For any inquiries, issues, or feedback, feel free to open an issue or contact us directly at nicola.melluso@kuleuven.be.
Part of this code is inspired from https://github.com/sam-arts/respol_patents_code
Users are reminded to adhere to copyright regulations and ethical guidelines when utilizing and adapting the provided resources and data.