Export
Research Overview
Wai Gen Yee focuses on information retrieval, web search optimization, and the analysis of user-generated content. Early work explored the utility of web user comments for improving search accuracy, particularly in multimedia contexts like video search. The trajectory of this research later shifted toward efficient, scalable methods for extracting aspect weights from text-based product reviews.
Research Trajectory
The available evidence outlines a research trajectory active primarily between 2009 and 2012, centered on leveraging user-generated content to enhance search and recommendation systems. A persistent challenge in information retrieval is the sparsity or coarseness of official metadata, which Yee addressed by mining user comments and reviews to enrich search indexes.
In 2009, Yee led an investigation into the foundational question of whether user feedback could improve retrieval systems, publishing Are Web User Comments Useful for Search?. This work established a baseline for using crowdsourced text to supplement traditional search algorithms. Building on this premise, Yee subsequently focused on the multimedia domain, where official textual descriptions are often too short to capture the precise content of a video. As first author, Yee introduced eduKEN: a tool for fine-grained video comment collection and analysis in 2010. This system was designed to extract supplemental information from user comments and apply it directly to search engine indexes, thereby improving video search accuracy. The implementation of such dynamic, comment-based indexing systems often intersects with infrastructural challenges, such as those found in the broader area of Asynchronous Background Maintenance for Real-Time Data Systems, as search engines must continuously integrate new user feedback without degrading query latency.
By 2012, Yee’s focus had expanded from search indexing to the semantic analysis of consumer reviews. Serving as senior author, Yee guided research on the Efficient estimation of aspect weights. This work addressed the computational bottleneck of modeling text in user reviews to determine how much consumers care about specific product aspects (such as a hotel's location, food, or amenities). The research introduced a linear-time method for computing these aspect weights, offering a significant scalability improvement over existing cubic-time state-of-the-art solutions.
Current Directions
Based on the provided corpus, the most recent paper in the evidence is from 2012. This later work indicates a shift toward highly scalable natural language processing and sentiment analysis within e-commerce and review platforms. The focus on reducing computational complexity—specifically moving from cubic to linear time algorithms for aspect weight estimation—suggests an overarching interest in making sophisticated text analytics viable for large-scale, real-world data systems.
Profile updated
| Title | Venue | Year |
|---|---|---|
| Efficient estimation of aspect weights | SIGIR | 2012 |
| eduKEN: a tool for fine-grained video comment collection and analysis | WWW | 2010 |
| Are Web User Comments Useful for Search? | LSDS-IR @ SIGIR | 2009 |
Loading new papers…
Loading research neighbors…
See something wrong? Report a mistake