1、Learning To Rank在个性化电商搜索中的应用 吴晨 (搜索 BG: Natural Artificial Intelligence) 2016.10.22 Background Learning to Rank Personalized E-Commerce Search Summary Reference 2 Outline Predict relevance scores and re-rank products returned by an e-commerce search engine on the search engine result page (SERP) Dat
2、a Using Search, browsing, and transac2on histories for all users and specically the user interac2ng with the search engine in the current session Product proper2es and meta-data Method Using Machine Learning (e.g. RankSVM, LambdaMart) Ranking Func2on (e.g. BM25, Cosine Similarity) 3 Background LEARN
3、ING TO RANK 4 Ranking Problem Learning to Match? Methods Pointwise Pairwise Listwise Theory (PAC) Generalization Stability Applications Search Recommender System Question Answering Sentiment Analysis 5 Introduc:on Machine Learning Supervised learning with labeled data Ranking of objects by subject F
4、eature based ranking function Approach Traditional BM25 (Probabilistic Model) New Query and associated products form Group (Train Data) Groups are i.i.d Features (query and product) in Group are not i.i.d Model is a function of features 6 Formula:on Data Labeling Relevance metric (Point) Ordered pai
5、rs Ordered list Feature Extraction Relevance (User/Query-Prod Feature) Semantic (User/Query-Prod Feature) Importance (Prod Feature) Learning Method Model Loss Function Optimized Algorithms Evaluation Measure NDCGk 7 Issues Machine Learning Classification Regression Ordinal Classification/Regression
6、Ordinal Regression Pointwise Transfer ranking to regression Ignore group info Learning to Rank Pairwise Transfer ranking to binar y classification Listwise Straightforward represent learning 8 Methods McRank (2007) Ordinal Liner Regression (Staged) Logistic Regression 9 Pointwise Model RankSVM (2000) Pairwise classification IR SVM (2006) Cost-sensitive Pairwise Using modified hinge loss RankBoost (2003) RankNet (2005) LambdaMart (2008) 10 Pairwise Model
