Amazon Technologies, Inc.

Xerox Corporation

Gregory Alan Hanneman

Alon Lavie

Gavin Richard Jewell

William Joseph Kaper

Alexandre Alexandrovich Klementiev

Michael Denkowski

Matthew Ryan Fiorillo

Robert Thomas Olszewski

Christopher Pitstick

Adrien Hubert Marie Pierard

Marc Dymetman

Nicola Cancedda

Ehud Hershkovich

Laura Josephine Kieras

Austin Matthews

Christopher James Dyer

Gyorgy Istvan Szarvas

Kartik Goyal

Technologies are disclosed for utilizing artificial intelligence-based machine translation to augment document summarization. Text can be extracted from a document in a first language. Machine translation can be utilized to translate the text from the first language to a second language. The translated text can be used to identify documents in the second language that include support for the translated text. A user interface can be provided that indicates the number of documents in the second language that provide support for the extracted text. Documents in the first language can also be translated to the second language. Documents that provide support for a text string can be identified in the documents translated to the second language and in other documents in the second language. A user interface can be provided that indicates the number of documents in the first language and the second language that provide support for the text.

Utilizing artificial intelligence-based machine translation to augment document summarization

Technologies are disclosed herein for statistical machine translation. In particular, the disclosed technologies include extensions to conventional machine translation pipelines: the use of multiple domain-specific and non-domain-specific dynamic language translation models and language models; cluster-based language models; and large-scale discriminative training. Incremental update technologies are also disclosed for use in updating a machine translation system in four areas: word alignment; translation modeling; language modeling; and parameter estimation. A mechanism is also disclosed for training and utilizing a runtime machine translation quality classifier for estimating the quality of machine translations without the benefit of reference translations. The runtime machine translation quality classifier is generated in a manner to offset imbalances in the number of training instances in various classes, and to assign a greater penalty to the misclassification of lower-quality translations as higher-quality translations than to misclassification of higher-quality translations as lower-quality translations.

Optimized statistical machine translation system with rapid adaptation capability

A machine translation system capable of incremental and preemptive machine translation is disclosed. Content items on a page can be provided to multiple machine translation services for translation. Each of the machine translation services is capable of translating content items at a different quality level. Content items translated at a lower quality level might be received before content items translated at a higher quality level and presented in a user interface (UI). When content items translated at a higher quality level are received, the translated content items can be used to replace the lower quality level translations previously presented in the UI. Content items referenced by pages identified in search results can also be preemptively machine translated, cached, and provided when requests are received for the translated content items.

Incremental and preemptive machine translation

Preemptive machine translation

A system and method for generating word alignments from pairs of aligned text strings are provided. A corpus of text strings provides pairs of text strings, primarily sentences, in source and target languages. A first alignment between a text string pair creates links therebetween. Each link links a single token of the first text string to a single token of the second text string. A second alignment also creates links between the text string pair. In some cases, these links may correspond to bi-phrases. A modified first alignment is generated by selectively modifying links in the first alignment which include a word which is infrequent in the corpus, based on links generated in the second alignment. This results in removing at least some of the links for the infrequent words, allowing more compact and better quality bi-phrases, with higher vocabulary coverage, to be extracted for use in a machine translation system.

Word alignment method and system for improved vocabulary coverage in statistical machine translation

Growing community of inventors

Pittsburgh, PA, United States of America

Gregory Alan Hanneman