JIAN WANG Mountain View, CA 94043 (650) 868-6572 jwang30@soe.ucsc.edu RESEARCH INTEREST Big Data Analysis, Recommender Systems, Data Mining, Personalization, Information Retrieval, Machine Learning, Large-scale systems EDUCATION University of California, Santa Cruz Santa Cruz, CA Ph.D. in Computer Science Sep. 2009 Jun. 2013 Lehigh University Bethlehem, PA M.Sc. in Computer Science Sep. 2007 - May 2009 Fudan University Shanghai, China B. E. in Communications Science and Engineering Sep. 2003 - Jun. 2007 EMPLOYMENT HISTORY LinkedIn, Mountain View, CA Senior Applied Research Engineer in RecLS team Jun. 2012 present Perform research and develop recommendation models in the job domain. The proposed recommender systems and models power the "jobs you may be interested in" product, which is LinkedIn's premier job recommendation product. It serves fresh job recommendations for more than 200 million LinkedIn members. The product accounts for more than 50% job applications in LinkedIn, which is one of LinkedIn s major revenue sources. Develop and optimize models for the big data analysis. Implement the model in a large-scale distributed system. Experience with a host of machine learning techniques for building real-world, scalable and game changing data products. Build the large-scale recommender system with millions of training data on Hadoop. University of California Santa Cruz, Computer Science Department, Santa Cruz, CA Research Assistant in IRKM lab Aug. 2009 Jun. 2013 Performed research in support of projects in the UCSC Information Retrieval and Knowledge Management laboratory under the direction of Professor Yi Zhang. Performed research of recommender systems in the e-commerce domain. Developed a prototype of the social music recommender system (www.fmvilla.com) with distributed system. Page 1
ebay Inc, San Jose, CA Research Intern in ebay Research labs Jun. 2010 Sep. 2010 Worked in Merchandising and Catalog Team, ebay research lab. Performed research on the post-purchase recommendation problem in the e-commerce website, with my mentor Neel Sundaresan and Badrul Sarwar. Developed a prototype of the proposed product-level post-purchase recommender system with the real-world e-commerce data. Published in RecSys 2011 and won the best short paper award. Lehigh University, Computer Science and Engineering Department, Bethlehem, PA Research Assistant in WUME lab Sep. 2007 Jul. 2009 Performed research, developed software, and prepared data in support of projects in the Lehigh Web Understanding, Modeling, and Evaluation laboratory under the direction of Professor Brian Davison. Completed installing and configuring Hadoop cluster on multiple clusters. PUBLICATION (Select papers in bold) Wang, J. and Hardtke, D. (2014) User Latent Preference Model for Better Downside Management in Recommender Systems. Submitted to WWW 14 (Full paper) Wang, J. and Zhang, Y.(2013) Opportunity Model for E-commerce Recommendation: Right Product, Right Time. In Proceedings of the 36 th International ACM Conference on Research and Development in Information Retrieval (SIGIR '13), Dublin, Ireland (Full paper, 19.9% Acceptance rate) Wang, J., Zhang, Y., Posse, C. and Bhasin, A.(2013) Is It Time for a Career Switch? In Proceedings of the 23 rd International World-Wide Web Conference (WWW 2013), Rio de Janeiro, Brazil (Full paper, 15% Acceptance rate) Wang, J., Zhang, Y. and Chen, T.(2012). Unified Recommendation and Search in E-commerce. In Proceedings of the 8 th Asia Information Retrieval Societies Conference (AIRS '12), TianJin, China (Short Paper, acceptance rate 35.1%) Wang, J. and Zhang, Y.(2011). Utilizing Marginal Net Utility for Recommendation in E-commerce. In Proceedings of the 34 th International ACM Conference on Research and Development in Information Retrieval (SIGIR '11), BeiJing, China (Full Paper, acceptance rate 19.8%) Wang, J., Sarwar, B.M. and Sundaresan, N.(2011). Utilizing Related Product for Post-Purchase Recommendation in E-commerce. In Proceedings of the 5 th ACM Conference on Recommender Systems (RecSys 2011). Chicago, USA (Best short paper award) (Short Paper, acceptance rate 40.7%) Tyler, S., Wang, J. and Zhang, Y.(2010). Utilizing Refinding for Personalized Information Retrieval. In Proceedings of the 19 th ACM Conference on Information and Knowledge Management (CIKM 2010), Toronto, Canada (Short Paper, acceptance rate 17.9%) Page 2
Wang, J., Hong, L. and Davison, B. (2009). RSDC 09: Tag Recommendation Using Keywords and Association Rules. In Proceedings of the ECML PKDD Discovery Challenge Workshop, Bled, Slovenia (Full Paper) Wang, J. and Davison, B. (2009). Counting Ancestors to Estimate Authority. In Proceedings of the 32 nd International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR '09), Boston, USA (Short Paper, acceptance rate 34%) Wang, J. and Davison, B. (2008). Explorations in Tag Suggestion and Query Expansion. In Proceedings of the Workshop on Search in Social Media (SSM 2008) at the 17 th ACM Conference on Information and Knowledge Management (CIKM 2008), Napa Valley, CA, USA (Full Paper) RESEARCH EXPERIENCE USER LATENT PREFERENCE MODEL FOR BETTER DOWNSIDE MANAGEMENT IN RECOMMENDER SYSTEMS Fall 2014 PI: Jian Wang LinkedIn Corp Downside management is an important topic in the field of recommender systems. User satisfaction increases when good items are recommended, but satisfaction drops significantly when bad recommendations are pushed to them. For example, a parent would be disappointed if violent movies are recommended to their kids and may stop using the recommendation system entirely. A vegetarian would feel steak-house recommendations useless. A CEO in a mid-sized company would feel offended by receiving intern-level job recommendations. Under circumstances where there is penalty for a bad recommendation, a bad recommendation is worse than no recommendation at all. While most existing work focuses on upside management (recommending the best items to users), this paper emphasizes achieving better downside management (reducing the recommendation of irrelevant or offensive items to users). The approach we propose is general and can be applied to any scenario or domain where downside management is key to the system. To tackle the problem, we design a user latent preference model to predict the user preference in a specific dimension, say, the dietary restrictions of the user, the acceptable level of adult content in a movie, or the geographical preference of a job seeker. We propose to use multinomial regression as the core model and extend it with a hierarchical Bayesian framework to address the problem of data sparsity. After the user latent preference is predicted, we leverage it to filter out downside items. We validate the soundness of our approach by evaluating it with an anonymous job application dataset on LinkedIn. The effectiveness of the latent preference model was demonstrated in both offline experiments and online A/B testings. The user latent preference model helps to improve the VPI (views per impression) and API (applications per impression) significantly which in turn achieves a higher user satisfaction. SESSION-AWARE RECOMMENDER SYSTEM IN E-COMMERCE Fall 2010 Spring 2013 PI: Yi Zhang The Information Retrieval and Knowledge Management Lab of UC Santa Cruz (IRKM) To enhance the consumer's experience, we propose to investigate the session-aware recommender systems in e-commerce sites. The product recommendation is viewed as a session-based Page 3
interactive process between the system and the user. We first explore how to integrate the complementary information within a single session to build a unified recommender system. To go beyond making recommendations within a single session, we then study how to make better recommendations across sessions. To make recommendations based on a user s previous behavior in earlier sessions, we need to understand how users make purchase decisions across sessions. To further incorporate the time interval between sessions into the system, we adapt the proportional hazards model in survival analysis and propose the new opportunity model in e-commerce. To our best knowledge, it would be the first step to analyze recommender systems in different stages within a session in research community. In addition, such session-aware systems can help the real world e-commerce site to better understand how user's preference changes within a session. OPPORTUNITY MODEL IN E-COMMERCE RECOMMENDATION Spring 2013 PI: Yi Zhang The Information Retrieval and Knowledge Management Lab of UC Santa Cruz (IRKM) We propose and study the new problem: how to recommend the right product at the right time? To solve this problem, we propose a principled approach, (i.e. the opportunity model), to predict the joint probability of purchasing a product and the time of the event. We extend the proportional hazards model in statistics with the hierarchical Bayesian framework as part of the solution, and derive detailed inference steps based on the variational Bayesian algorithm. We leverage the joint probability in both the zero-query pull-based recommendation scenario and the proactive push-based email/message promotion scenario. In particular, the probability enables a proactive recommendation agent to decide whether to send recommendations of certain items to a user at a particular time based on a solid utility optimization framework. Experimental results show that the opportunity modeling approach significantly improves the user satisfaction and the conversion rate of the system. RECOMMENDING THE RIGHT JOB AT THE RIGHT TIME Fall 2012 LinkedIn Corp PI: Yi Zhang, Anmol Bhasin Tenure is a critical factor for an individual to consider when making a job transition. For instance, software engineers make a job transition to senior software engineers in a span of 2 years on average, or it takes for approximately 3 years for realtors to switch to brokers. While most existing work on recommender systems focuses on finding what to recommend to a user, this project places emphasis on when to make appropriate recommendations and its impact on the item selection in the context of a job recommender system. Our approach is inspired by the proportional hazards model in statistics. It models the tenure between two successive decisions and related factors. We further extend the model with a hierarchical Bayesian framework to address the problem of data sparsity. The proposed model estimates the likelihood of a user's decision to make a job transition at a certain time, which is denoted as the tenure-based decision probability. New and appropriate evaluation metrics are designed to analyze the model's performance on deciding when is the right time to recommend a job to a user. We validate the soundness of our approach by evaluating it with a real-world job application dataset across 140+ industries on LinkedIn. It contains millions of job applications from millions of users across several months. Experimental results show that the hierarchical proportional hazards model has better predictability of the user's decision time, which in turn helps the recommender system to achieve higher utility/user satisfaction. Page 4
UNIFIED RECOMMENDATION AND SEARCH IN E-COMMERCE Spring 2012 PI: Yi Zhang The Information Retrieval and Knowledge Management Lab of UC Santa Cruz (IRKM) This project explores how to integrate the complementary information to build a unified recommendation and search system. We propose a new three-level graphical model as the unified model to better understand the user's purchase intention. It explicitly models the user's categorical choice, purchase state (repurchase, variety seeking or new purchase) in addition to the final product choice. Experiments on a data from an e-commerce website (shop.com) show that the unified model works better than the basic search or recommendation systems on average, particularly for the repeated purchase situations. In addition, the graphical model predicts a user's categorical choice and purchase state reasonably well. The insight and predicted purchase state may be useful for implementing the user-state specific marketing and advertising strategies. UTILIZING MARGINAL NET UTILITY FOR RECOMMENDATION IN E-COMMERCE Fall 2011 PI: Yi Zhang The Information Retrieval and Knowledge Management Lab of UC Santa Cruz (IRKM) Traditional recommendation algorithms often select products with the highest predicted ratings to recommend. However, earlier research in economics and marketing indicates that a consumer usually makes purchase decision(s) based on the product's marginal net utility (i.e., the marginal utility minus the product price. To better match users' purchase decisions in the real world, this paper explored how to recommend products with the highest marginal net utility in e-commerce sites. Inspired by the Cobb-Douglas utility function in consumer behavior theory, we proposed a novel utility-based recommendation framework. The framework could be utilized to revamp a family of existing recommendation algorithms. To demonstrate the idea, we used Singular Value Decomposition (SVD) as an example and revamped it with the framework. We evaluated the proposed algorithm on an e-commerce (shop.com) data set. The new algorithm significantly improved the base algorithm, largely due to its ability to recommend both products that are new to the user and products that the user is likely to re-purchase. POST-PURCHASE RECOMMENDATION, EBAY INC. Summer 2010 ebay Research Labs PI: Neel Sundaresan In this project, we design a recommender system for the post-purchase stage, i.e., after a user purchases a product. Our method combines both behavioral and content aspects of recommendations. We first find the most related categories for the active product in the post-purchase stage. Among these related categories, products with high behavioral relevance and content relevance are recommended to the user. In addition, our algorithm considers the temporal factor, i.e., the purchase time of the active product and the recommendation time. We apply our algorithm on a real-world purchase data from ebay. Comparing to the baseline item-based collaborative filtering approach, our hybrid recommender system achieves significant coverage and purchase rate gain for different time windows. Page 5
PROFESSIONAL ACTIVITIES Journal Reviewer: Invited Reviewer of ACM Transactions on Interactive Intelligent Systems (TIIS), 2014 - present Invited Reviewer of Computational Intelligence, 2013 - present Invited Reviewer of World Wide Web Journal (WWWJ), 2013 present Invited Reviewer of Information Processing & Management (IPM), 2013 present Invited Reviewer of Information Retrieval Journal, 2013 present Invited Reviewer of Transactions on Intelligent Systems and Technology (TIST), 2013 present Invited Reviewer of IEEE Transactions on Cybernetics, 2013 - present Invited Reviewer of ACM Transactions on Knowledge and Data Engineering (TKDE), 2012 - present Invited Reviewer of ACM Transactions on Information Systems (TOIS), 2011 - present Organizing Committee Member: Local Organization Co-Chair of the 8 th ACM Recommender Systems (RecSys 2014) Co-chair of the 5 th Social Recommender Systems (SRS2014) at World-Wide Web Conference (WWW 2014) Program Committee Member: The 24 th International World-Wide Web Conference (WWW 2015) The 23 rd Conference on User Modeling, Adaption and Personalization, Full paper (UMAP 2015) The International Conference on Information and Knowledge Management (CIKM 2014) The 8th ACM Recommender System Conference (RecSys 2014) (Full paper, short paper, demo) The 20 th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (SIGKDD 2014) The 23 rd International World-Wide Web Conference (WWW 2014) The 37th ACM International Conference on Research and Development in Information Retrieval (SIGIR 2014) The International Conference on Multimedia Retrieval (ICMR 2014) The 22 nd Conference on User Modeling, Adaption and Personalization, Full paper (UMAP 2014) The 22 nd Conference on User Modeling, Adaption and Personalization, Poster and Demonstration (UMAP 2014) The 28 th AAAI Conference on Artificial Intelligence (AAAI 2014) The 8 th International Conference on Weblogs and Social Media (ICWSM 2014) The IEEE/ACM International Conference on Advances in Social Network Analysis and Mining (ASONAM 2014) The 7th IEEE International Conference on Social Computing and Networking (SocialCom2014) The 1 st, 2 nd Workshop on User Engagement Optimization (UEO2013, 2014) The 9 th Asian Information Retrieval Societies Conference (AIRS 2013) Conference Reviewer: ACM KDD 2008, 2013 ACM SIGIR 2008, 2009, 2010, 2011 ACM WWW 2008, 2009 ACM CIKM 2008, AIRWeb 2008, WSDM 2009, ICDM 2009, RecSys 2011, OAIR 2013 Page 6
TALKS AND PRESENTATIONS (Nov. 2014) Recommender Systems in LinkedIn. NYU Shanghai-Symposium on Data Science and Applications 2014, Shanghai, China. (Jul. 2013) Opportunity Model for E-commerce Recommendation: Right Product; Right Time. at 36 th ACM Conference on Research and Development in Information Retrieval (SIGIR 2013), Dublin, Ireland. (May 2013) Is It Time for a Career Switch? at 22 nd International World Wide Web Conference, Rio de Janeiro, Brazil (Feb. 2013) Recommendation in E-commerce Sites: Right product, Right time. at Shanghai UnionPay Smart Corp, Shanghai, China. (Dec. 2012) Unified Recommendation and Search in E-commerce at the 8 th Annual Asia Information Retrieval Societies Conference (AIRS 2012), TianJin, China. (Dec. 2012) When to Make the Right Recommendation? at the 2012 Frontiers of Information Science and Technology(FIST) Workshop, Shanghai, China. (Oct. 2012) Recommendation in the Job Domain. at the 4 th Annual SRL/ISSDM Research Symposium, Santa Cruz, USA. (Oct. 2011) Utilizing Related Product for Post-Purchase Recommendation in E-commerce. at the 5 th ACM Recommender System, Chicago, USA. [Best short paper award] (Oct. 2011) Utilizing Marginal Net Utility for Recommendation in E-commerce. at the 3 rd Annual SRL/ISSDM Research Symposium, Workshop on Knowledge Management: Analytics and Big Data, Santa Cruz, USA. (Aug. 2011) Utilizing Marginal Net Utility for Recommendation in E-commerce. at Baidu R&D center, Shanghai, China. (Jul. 2011) Utilizing Marginal Net Utility for Recommendation in E-commerce. at the 34 th ACM Conference on Research and Development in Information Retrieval (SIGIR 2011), Beijing, China. (Jul. 2009) Counting Ancestors to Estimate Authority at the 32 nd ACM Conference on Research and Development in Information Retrieval (SIGIR 2009), Boston, USA. (Oct. 2008) Explorations in Tag Suggestion and Query Expansion. at the CIKM 2008 Workshop on Search in Social Media (SSM 2008), Napa Valley, CA, USA TEACHING EXPERIENCE (Fall 2014) Guest lecturer, TIM 260: Information Retrieval (Spring 2013) Teaching Assistant, CS 182: Introduction to Database Management Systems (Spring 2011) Teaching Assistant, ISM 58: System Analysis and Design (Spring 2010) Teaching Assistant, ISM 58: System Analysis and Design HONORS AND AWARDS SIGIR 2013 Student Travel Grant from Google and Donald B. Crouch Travel Grant WWW 2013 Student Scholarship Best Short Paper Award of RecSys 2011 SIGIR 2011 Student Travel Grant and Donald B. Crouch Travel Grant SIGIR 2009 Student Travel Grant Regents' Fellowship, Department of Computer Science, UC Santa Cruz, 2009 Dean s Doctoral Student Assistantship, Lehigh University, 2007 People s Scholarship, Fudan University, 2003-2007 Honorable Mention in Mathematical Contest in Modeling (MCM), 2002, 2003 Page 7
REFERENCES Available upon request Page 8