Translation system adapted for query translation via a reranking framework
First Claim
Patent Images
1. A translation method comprising:
- receiving an input query in a source language; and
outputting a target query, the target query being identified from a set of candidate target queries, each target query being based on a translation of the input query into a target language, different from the source language, with a machine translation system which includes a reranking model for ranking the candidate target queries, which has been trained by a method which includes;
for each of a plurality of training queries in the source language, translating the training query in the source language into the target language to generate translated queries which are each a translation of the respective training query;
for each of the translated queries;
computing a feature representation of the translated query;
retrieving a set of annotated documents from a document collection in response to the translated query, the documents in the retrieved set of annotated documents including annotations that are based on responsiveness of each of the documents to each of the training queries, andcomputing a precision score for the translated query based on relevance scores of the retrieved documents in the set of annotated documents, each of the relevance scores being based on the annotations of the documents in the retrieved set of annotated documents; and
learning feature weights for the reranking model based on the precision scores and feature representations of the translated queries.
1 Assignment
0 Petitions
Accused Products
Abstract
A system and method suited to translation of queries are disclosed. The method includes receiving an input query in a source language and outputting a target query, based on a translation of the input query into a target language, different from the source language. The translation is performed with a machine translation system which has been trained on representations of features of translated queries that have been generated by translation of an original query, in the source language, into the target language and a measure of information retrieval performance of each the translated queries, for each of a set of original queries.
27 Citations
24 Claims
-
1. A translation method comprising:
-
receiving an input query in a source language; and outputting a target query, the target query being identified from a set of candidate target queries, each target query being based on a translation of the input query into a target language, different from the source language, with a machine translation system which includes a reranking model for ranking the candidate target queries, which has been trained by a method which includes; for each of a plurality of training queries in the source language, translating the training query in the source language into the target language to generate translated queries which are each a translation of the respective training query; for each of the translated queries; computing a feature representation of the translated query; retrieving a set of annotated documents from a document collection in response to the translated query, the documents in the retrieved set of annotated documents including annotations that are based on responsiveness of each of the documents to each of the training queries, and computing a precision score for the translated query based on relevance scores of the retrieved documents in the set of annotated documents, each of the relevance scores being based on the annotations of the documents in the retrieved set of annotated documents; and learning feature weights for the reranking model based on the precision scores and feature representations of the translated queries. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19)
-
-
20. A query translation system comprising:
-
a decoder which receives a source query in a source language and outputs a set of candidate queries in a target language, each of the candidate queries being a translation of the same source query; a reranking module which outputs a target query based on at least one of the candidate queries, the reranking module extracting features of each of the candidate queries and computing a function in which the extracted features are weighted by feature weights, the feature weights having been learned on features of each of a set of translated queries generated by translation of each of a set of training queries into the target language and a precision score computed for each of the translated queries, the computing of the precision for each of the translated queries including retrieving a set of annotated documents from a document collection in response to the translated query, the documents in the retrieved set of annotated documents including annotations that are based on responsiveness of each of the documents to each of the training queries, and computing the precision score for the translated query based on relevance scores of the retrieved documents in the set of annotated documents, each of the relevance scores being based on the annotations of the documents in the retrieved set of annotated documents; and a processor which implements the reranking module. - View Dependent Claims (21)
-
-
22. A method for training a translation system for translation of queries, comprising:
-
for each of a plurality of training queries in a source language; translating the training query to generate a set of translated queries in a target language; for each translated query in the set of translated queries; extracting values of features for each of a finite set of features; retrieving a set of annotated documents from a document collection in response to the translated query, the documents in the retrieved set of annotated documents including annotations that are based on responsiveness of each of the documents to each of the training queries; and computing a precision score for the translated query based on relevance scores of the retrieved documents in the set of annotated documents, each of the relevance scores being based on the annotations of the documents in the retrieved set of annotated documents; learning feature weights for each of the features based on the extracted values of the features and the respective precision score of each translated query; and storing the feature weights for use in translating a new query, different from each of the training queries, from the source language to the target language, whereby candidate translations of the new query are ranked based on their extracted values of features and the stored feature weights. - View Dependent Claims (23, 24)
-
Specification