resume
(176 Words, 1 Minutes)
Jessie Liu
- Tel: (+86)185 9611 5587
- Email: 201983030@uibe.edu.cn
EDUCATION
University of International Business and Economics, School of Information Technology & Management Bachelor of Engineering: Data Science and Big Data Technology, June 2023 Cumulative GPA: 3.41/4.00 Relevant Coursework: C++ Programming (4.0), Data Structure (3.7), JAVA Programming (3.7), Python and Big Data Analysis (4.0), Data Visualization and Applications (4.0), Information Retrieval and Data Processing (3.7)
PROJECTS
NLP-centric MultiModal Biological Experiment Assistant Institute of Automation, Chinese Academy of Sciences, Research Assistant September 2023 – Present Github Link: https://github.com/juanliu27/Langchain-Chatchat
- Customized a protocol generation system with LangChain framework, Chatchat model and used Chatglm-6b model and bge-reranker-large embedding model after assessment.•Crafted a JSON recognizer that maps 204 actions to tasks based on real-world protocols, featuring a duration and condition detection function. Utilized Few-Shot learning to design prompts, enhancing control over the text output direction.
- Constructed a Knowledge Base Question Answering System containing over 20,000 protocols sourced through web crawling. Transformed these files into markdown format for better handling, and leveraged faiss for vectorization. Additionally, incorporated keyword embedding to significantly improve the system’s recall rate.
Analysis of Virtual Anchor Profitability based on System Dynamics UIBE, Researcher November 2022 - May 2023 Github Link: https://github.com/juanliu27/Vtuber_Profitability
- Scraped 184,622 data covering live broadcasts, revenue, and videos of vtubers over past three years using Beautifulsoup.•Implemented an iterative refinement process, selectively filtering out variables with p-values exceeding 0.05 while analyzing the correlation between various characteristics and virtual monetization capabilities with OLS regression.
- Established a dynamic virtual liver profit estimating system using Vensim, with errors not surpassing 1%.
Prediction of Weibo Trends based on the Random Forest UIBE, Researcher June 2021 - December 2021 Github Link: https://github.com/juanliu27/Weibo-Trend-Predictions
- Analyzed dataset labels using Spearman correlation coefficients and visualized them using a heatmap. Chose labels with a correlation coefficient greater than 0.01 related to the ‘hot’ label.
- Trained a model using the random forest method on a dataset of over 31,700 Weibo entries. Fine-tuned n_estimators by analyzing the learning curve, leading to an improvement in the cross-validation score from 0.75 to 0.81