Log In Sign Up

Optimizing Query Evaluations using Reinforcement Learning for Web Search

by   Corby Rosset, et al.

In web search, typically a candidate generation step selects a small set of documents---from collections containing as many as billions of web pages---that are subsequently ranked and pruned before being presented to the user. In Bing, the candidate generation involves scanning the index using statically designed match plans that prescribe sequences of different match criteria and stopping conditions. In this work, we pose match planning as a reinforcement learning task and observe up to 20 degradation in the quality of the candidate sets.


page 1

page 2

page 3

page 4


Semantic Web Search based on Ontology Modeling using Protege Reasoner

The Semantic Web works on the existing Web which presents the meaning of...

Studying Ranking-Incentivized Web Dynamics

The ranking incentives of many authors of Web pages play an important ro...

GAP2WSS: A Genetic Algorithm based on the Pareto Principle for Web Service Selection

Despite all the progress in Web service selection, the need for an appro...

Unsupervised Common Question Generation from Multiple Documents using Reinforced Contrastive Coordinator

Web search engines today return a ranked list of document links in respo...

Evolution of a Web-Scale Near Duplicate Image Detection System

Detecting near duplicate images is fundamental to the content ecosystem ...

Heterotic String Model Building with Monad Bundles and Reinforcement Learning

We use reinforcement learning as a means of constructing string compacti...

Roomsemble: Progressive web application for intuitive property search

A successful real estate search process involves locating a property tha...