Efficient label acquisition for query rewriting

摘要

Systems, methods, and computer storage media having computer-executable instructions embodied thereon for rewriting queries and labeling word pairs. Queries are received and alternate words are identified for word pairs (i.e., query words and alternate words). Word pair links are presented to users and indicators are received based on actions taken by the users. Labels are assigned to the word pairs based on the indicators and communicated to a classifier.

主权项

1. A method of determining which alternate word to include in a word pair link, the method comprising:
receiving, at a computing device, an initial query comprising a query word; identifying a first alternate query comprising a first alternate word, wherein the first alternate word comprises more than one element; determining an initial term frequency of the query word in snippets of results for the initial query; determining a first alternate term frequency for each element of the first alternate word in snippets of results for the initial query; comparing the initial term frequency to the first alternate term frequency for each element of the first alternate word; assigning a blame to each element of the first alternate word based on comparing term frequencies, wherein the blame indicates an amount of change in term frequency caused by each element of the first alternate word; determining the first alternate term frequency satisfies a configurable threshold; displaying a word pair link comprising the query word and the first alternate word, wherein the word pair link prompts a user to select the query word or the first alternate word; receiving indicators for the word pair; and assigning a label to the word pair based on the indicators indicating whether the query word and the first alternate word are synonyms.