Jinhao L.

Key Skills

Software

Other

Top Subject Matter

NLP sentiment analysis and market sentiment monitoring

LLM-assisted information extraction and semantic data cleansing for recommendations

Top Data Types

Text

Top Task Types

Emotion Recognition

Text Generation

Data Collection

Freelancer Overview

Sentiment Analysis System Using LSTM + CNN. Brings 1+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Other. Education includes Master of Computer Science, The University of Sydney (2025) and Bachelor of Mechanical Engineering, Nanchang University (2023). AI-training focus includes data types such as Text and labeling workflows including Emotion Recognition and Text Generation.

Labeling Experience

LLM-assisted Data Extraction and Semantic Cleansing Pipeline

OtherTextText Generation

Built an LLM-assisted data extraction pipeline to generate structured recommendation data from Reddit discussions and Amazon product metadata. The workflow included fuzzy matching to resolve sparse explicit product links and semantic cleansing to validate and align outputs. • Scaled data acquisition across 29 subreddits and 6 product categories using URL extraction and ASIN parsing • Applied LLM fuzzy matching when product links were sparse to increase usable samples • Designed semantic cleansing (confirmation, sensitive-word filtering, metadata alignment, and LLM/SBERT validation) • Standardized price mapping and improved metadata coverage while reducing re-run time

2025 - 2025

Sentiment Analysis System Using LSTM + CNN

OtherTextEmotion Recognition

Built an end-to-end NLP sentiment classification pipeline using Sentiment140 tweets for user review analysis and market sentiment monitoring. Focused on preparing and transforming raw text into model-ready inputs by performing lexical annotation and feature extraction. The work included tuning and evaluating the model to improve classification quality and interpretability. • Performed text preprocessing (regex noise removal, tokenization, stop-word filtering, stemming) • Conducted lexical annotation and feature extraction for sentiment prediction • Implemented a hybrid LSTM + CNN architecture for context and n-gram feature modeling • Used Bayesian optimization for hyperparameters and performed error analysis on misclassified cases

2024 - 2024

Education

Work History