Microsoft Technology Licensing, LLC

Anthony Aue

Michael Gamon

Arul A Menezes

Eric K Ringger

Christopher Brian Quirk

Simon H Corston-Oliver

Robert C Moore

Frederik Schaffalitzky

Ibrahim Eden

Jonas Lindblom

Michel Galley

Pierre P N Greborio

Fredrik Furesjö

Fredrik Furesjo

Pierre Pn Greborio

Jonas Nils Lindblom

Pierre PN Greborio

The disclosure pertains to a communication system for effecting a voice or video call between at least a source user speaking a source language and a target user speaking a target language. A translation procedure is performed on call audio of the call to generate an audio translation of the source user's speech in the target language for outputting to the target user. A notification is outputted to the target user to notify the target user of a change in the behavior of the translation procedure, the change relating to the generation of the translation.

In-call translation

The claimed subject matter provides a system and/or method for segmenting a multi-language text. An exemplary method comprises determining an initial probability distribution for sentences in the multi-language text, the initial probability distribution indicating the likelihood of each sentence being in each of a set of languages. A probability of language transitions across sentences may be learned based on the initial probability distribution. Additionally, a highest probability language sequence of sentences in the multi-language text may be determined based on a combination of the probability of language transitions and the prior probability distribution provided by an initial model. Further, web documents are annotated at a sentence by sentence level such that each sentence of a web document is labeled in a given language according to the highest probability language determined.

Language segmentation of multilingual texts

The described implementations relate to processing of electronic data. One implementation is manifested as a system that can include a cache module and at least one processing device configured to execute the cache module. The cache module can be configured to store data items in slots of a cache structure, receive a request for an individual data item that maps to an individual slot of the cache structure, and, when the individual slot of the cache structure is not available, return without further processing the request. For example, the request can be received from a calling application or thread that can proceed without blocking irrespective of whether the request is fulfilled by the cache module.

Non-blocking caching technique

Automated language translation often involves language translation resources of significant size (e.g., 50-gigabyte phrase tables) and significant computational power exceeding the capabilities of many mobile devices. Remotely accessible servers capable of near-realtime, automated translation may be inaccessible or prohibitively costly while traveling abroad. Presented herein are adaptations of language translation techniques for offline mobile devices involving reducing the size and raising the efficiency of the language modeling resources. A word index may be provided that stores respective string representations of the words of a language, and maps respective words to a location (e.g., address or offset) of respective word representations within the word index. Language translation resources (e.g., phrase tables) may then specify logical relationships using the word index addresses of the involved words, rather than the string equivalents. This technique significantly condenses the language resources and provides faster, bidirectional access to the word representations of the language.

Language processing resources for automated mobile language translation

A system and method for segmenting a multi-language text is provided. An exemplary method comprises determining an initial probability distribution for sentences in the multi-language text, the initial probability distribution indicating the likelihood of each sentence being in each of a set of languages. A probability of language transitions across sentences may be learned based on the initial probability distribution. Additionally, a highest probability language sequence of sentences in the multi-language text may be determined based on a combination of the probability of language transitions and the prior probability distribution provided by an initial model.

Technology is described for identifying language translations for source documents. The method includes finding source documents containing links to target documents and the link anchors of the links have language indicating text. A first tuple set can be generated for paired source documents and target documents with an expected target language for a target document. The first tuple set can be annotated with primary languages for the source documents and target documents to form a second tuple set where primary languages of the source documents and target documents are different. Further, a third tuple set can be generated using the second tuple set using a count of the number of times source documents and target documents occur in the first tuple set. Tuples can be removed from the third tuple set where a count ratio between source document count and target document count is less than a reference ratio.

Identifying language translations for source documents using links

A computer-implemented system and method for assessing the editorial quality of a textual unit (document, paragraph or sentence) is provided. The method includes generating a plurality of training-time feature vectors by automatically extracting features from first and last versions of training documents. The method also includes training a machine-learned classifier based on the plurality of training-time feature vectors. A run-time feature vector is generated for the textual unit to be assessed by automatically extracting features from the textual unit. The run-time feature vector is evaluated using the machine-learned classifier to provide an assessment of the editorial quality of the textual unit.

Technique for document editorial quality assessment

The present invention provides a system for identifying, extracting, clustering and analyzing sentiment-bearing text. In one embodiment, the invention implements a pipeline capable of accessing raw text and presenting it in a highly usable and intuitive way.

Method and apparatus for processing sentiment-bearing text

System for processing sentiment-bearing text

A method of decoding an input semantic structure to generate an output semantic structure. A set of transfer mappings are provided. A score is calculated for at least one transfer mapping in the set of transfer mappings using a statistical model. At least one transfer mapping is selected based on the score and used to construct the output semantic structure.

Growing community of inventors

Seattle, WA, United States of America

Anthony Aue