E2-477, UC Santa Cruz
1156 High Street, Santa Cruz, CA 95064
I'm now working at Twitter on ads CTR prediction - an awesome project!
I was a PhD candidate of the Information Retrieval and Knowledge Management (IRKM) Lab at University of California, Santa Cruz. My Advisor was Professor Yi Zhang. You can find my CV here.
Defended my dissertation, now officially a doctor!
Specific topics I have worked on: Personalized Information Filtering, Text Filtering/Mining, Recommender System, Search Results Summarization, Social Media Analysis, etc.
General interests: Applied Machine Learning, Data Mining, Statistics
I am passionate in using machine learning techniques to solve realistic problems in various domains and applications.
1) June - September, 2012. Research Intern at Microsoft Research Redmond
2) June - September, 2011. Research Intern at IBM Research Almaden Research Center
Lanbo Zhang. Content-Based Filtering for Semi-Structured Documents. PhD dissertation.
Lanbo Zhang, Yi Zhang, Yunfei Chen. Summarizing Highly Structured Documents For Effective Search Interaction. In Proceedings of the 35th ACM SIGIR Conference, 2012 (SIGIR 2012). Acceptance rate: 20%.
Structured data like products, movies, etc. are very common types of information people are searching every day. Search results snippets of structured data should be informative enough so that users can make right decisions on whether one result is relevant or not. In this paper, we study the problem of generating good summaries (snippets) for search results of structured data. The proposed method is evaluated using a game we designed on Amazon Mechanical Turk. See our paper for details.
Yunfei Chen, Lanbo Zhang, Aaron Michelony, Yi Zhang. 4Is of Social Bully Filtering: Identity, Inference, Influence and Intervention. In Proceedings of the 21th ACM CIKM conference, 2012 (CIKM 2012). Demo Paper.
Lanbo Zhang, Yi Zhang, Qianli Xing. Filtering Semi-Structured Documents Based on Faceted Feedback. In Proceedings of the 34th ACM SIGIR Conference, 2011 (SIGIR 2011). Full paper. Acceptance rate: 19.8%
A machine learner can not only learn from labeled instances, but labeled features. Compared with terms, faceted features (such as, Director: James Cameron, Actors: Tom Cruise) usually convey more information and are far less ambiguous. Users might be able to provide more accurate labels on faceted features. In this paper, we study the problem of how a machine learner can learn from labeled faceted features in the application of document filtering. See our paper for details.
Qianli Xing, Yi Zhang, Lanbo Zhang. On Bias Problem in Relevance Feedback. In Proceedings of the 20th ACM CIKM conference, 2011 (CIKM 2011). Poster.
Lanbo Zhang, Yi Zhang. Interactive Retrieval based on Faceted Feedback. In Proceedings of the 33rd ACM SIGIR Conference, 2010 (SIGIR 2010). Full paper. Acceptance rate: 16.7%
Compared with terms, document facets (such as, Language, Topic, Category, usually stored in metadata) usually convey more information about a document and are far less ambiguous. Searchers might be able to provide accurate feedback on document facets. In this paper, we study the problem of automatically presenting faceted features for users to interact with in the scenario of semi-structured documents retrieval. See our paper for details.
Lanbo Zhang, Yi Zhang. Discriminative Factored Prior Model for Personalized Content-Based Recommendation. In Proceedings of the 19th ACM CIKM Conference, 2010 (CIKM 2010).
Lanbo Zhang, Yi Zhang, Jadiel de Arma, Kai Yu. UCSC at Relevance Feedback Track. In Proceedings of the 17th Text REtrieval Conference (TREC), Gaithersburg, USA, 2009.
Lanbo Zhang, Shancong Zhang, Weiwei Chen. A Dynamic Link Model of Code on Embedded System. Computer Engineering & Design (in Chinese). 2008
Lanbo Zhang, Xiaomin Jin. Design of a Web-based Search Engine for Academic Papers. Technical Report (in Chinese), 2005
Lanbo Zhang, Xiaomin Jin. Implementation of Text Watermark System Based on Synonym Replacement. Technical Report (in Chinese), 2004
1) Providing organized content. MS 338023.01 (pending). We invented a new way to organize a collection of documents that maps a set of documents on a particular topic onto a spine document.
1) PC member, SIGIR 2013
2) PC member IJCLNP 2013
3) PC member CIKM 2013
4) PC member, ECIR 2012
5) PC member, CIKM 2012 (Poster)
6) PC member, SIGIR 2012 (Poster)
7) PC member, ECIR 2011
8) PC member, AIRS 2011
9) Reviewer for World Wide Web Journal
10) Reviewer for IJCNLP 2011, AIRS 2010, WWWJ, etc.