Chemistry Paper Mining: Reaction Information Extraction

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering

Final Year Thesis Oral Defense

Title: "Chemistry Paper Mining: Reaction Information Extraction"

By

HU Yicheng

Abstract:

Chemical reactions serve as the core of chemistry academic research, and 
they usually serve as the core contributions or central focuses of 
chemistry research publications. It is therefore essential for chemistry 
researchers, regardless of being an expert or a novice, to quickly grasp 
the reactions that appeared in the publications; or for data scientists to 
efficiently perform large-scale reaction extractions. These processes are 
largely manual and time-consuming despite the great significance, which 
creates a headache for the abovementioned people.

After extensive surveys and to respond to the demand of the chemistry 
research area, this project focuses on the extraction of graphical 
reaction representations within chemistry publications, using a 
heuristic-AI-based approach. It identifies common graphical patterns and 
utilizes them in recognizing and differentiating different components of 
chemical reactions, including the reactants, products, and reaction 
conditions; passing the identified chemical structures for AI-based 
recognition, that builds upon the vision transformers; and assembling the 
pool of processed information into the form that facilitates future works 
of chemistry researchers and data scientists.


Date            : 3 May 2022 (Tuesday)

Time            : 14:00-14:40

Zoom Link:
https://hkust.zoom.us/j/99433676483?pwd=c1hubVFnNXFjNzVla0F4Y25wbUpVUT09

Meeting ID      : 994 3367 6483

Passcode        : 610124

Advisor         : Prof. LUO Qiong

2nd Reader      : Dr. MA Xiaojuan