Match repair to remove matching documents

Abstract

The techniques described herein provide a match repair phase that removes matching documents that are identified for a search query as not actually containing items from the search query. A representation of each document (e.g., a forward index storing a list of items for each document) is used to identify valid matching documents (i.e., documents containing items from the search query) and invalid matching documents (i.e., not containing items from Documentation of the item being searched for). Any invalid matching documents are removed from further processing and ranking for the search query.

Description

Background technique [0001] The amount of information and digital content available on the Internet and other electronic resources continues to grow rapidly. In view of the vast amount of information, search engines have been developed to facilitate searching electronic documents. In particular, a user or computer may search for information and documents by submitting a search query, which may include, for example, one or more words. After receiving a search query, the search engine identifies relevant documents based on the search query. [0002] At a high level, search engines identify search results by ranking documents by their relevance to the search query. Ranking is usually based on a large number of document features. Given the large collection of documents, it is not feasible to rank all documents for a search query as it would take unacceptable time. Accordingly, search engines typically employ pipelines that include preliminary operations to remove documents fro...

AI Technical Summary

Problems solved by technology

Given the large collection of documents, it is infeasible to rank all documents for a search query as it would take unacceptable time

Images