Assessing contributions of scientific articles using transformer-based deep learning

M.Sc. Student: Qi He | Advisor: Jimmy Chih-Hsien Peng | Project Duration: 2023-2024

The number of published scientific articles is increasing at an exponential rate per year. Their claimed contributions may overlap to a certain extent, which is difficult to identify until the reader has reviewed all the works. To optimise the effort in conducting a literature review for research, e.g., in the field of power engineering, there is a need to develop a tool that can quantitatively assess the contributions of scientific articles. Leveraging the advancements in transformer-based learning algorithms, this project seeks to develop a tool that will evaluate the quality of a given article.

Quickly Start

Preparation

OpenAI api key by following this link: How to get an Openai api keys?

Nougat api path can be installed using this link: install Nougat

(Recommend using pytorch GPU environment)

Visit the website

Click here to visit !

Settings

Fill in your Openai api key and nougat path.
Click Save button.

Start to use

Upload your PDF paper files.
Wait for the analysis to complete.
Refer to demonstration at the top of page.