Abstract: Systems and methods for efficient online domain adaptation are provided herein. Methods may include receiving a post-edited machine translated sentence pair, updating a machine translation model by adjusting translation weights for a translation memory and a language model while generating test machine translations of the machine translated sentence pair until one of the test machine translations approximately matches the post-edits for the machine translated sentence pair, and retranslating the remaining machine translation sentence pairs that have yet to be post-edited using the updated machine translation model.

Abstract: Personalizing machine translation via online adaptation is described herein. According to some embodiments, methods for providing personalized machine translations may include receiving translator feedback regarding machine translations generated by a machine translation system for a translator, determining translator feedback that improves translations generated by the machine translation system, and incorporating the determined translator feedback into the translation methodology of the machine translation system to personalize the translation methodology.

Abstract: A method, system, and computer readable medium for translating a document is provided. A statistical machine translation engine is trained using a translation memory comprising an annotation. A translation rule associated with the annotation is defined. A source document in a source language is received. The source document comprises an instance of the annotation and a string. The string is translated using the statistical machine translation engine. The instance of the annotation is processed according to the translation rule. A target document in a target language is generated based on the translated string and the processed annotation.

Abstract: Systems and methods for efficient online domain adaptation are provided herein. Methods may include receiving a post-edited machine translated sentence pair, updating a machine translation model by adjusting translation weights for a translation memory and a language model while generating test machine translations of the machine translated sentence pair until one of the test machine translations approximately matches the post-edits for the machine translated sentence pair, and retranslating the remaining machine translation sentence pairs that have yet to be post-edited using the updated machine translation model.

Abstract: A document containing text in a source language may be translated into a target language based on content associated with that document, in conjunction with the present technology. An indication to perform an optimal translation of a document into a target language may be received via a user interface. The document may then be accessed by a computing device. The optimal translation is executed by a preferred translation engine of a plurality of available translation engines. The preferred translation engine is the most likely to produce the most accurate translation of the document among the plurality of available translation engines. Additionally, the preferred translation engine may be identified based on content associated with the document. The document is translated into the target language using the preferred translation engine to obtain a translated document, which may then be outputted by a computing device.

Abstract: The present technology may translate a content of a web page such as content locator (e.g., a uniform resource locator (URL)) from a source language to a target language. The content locator may be associated with a content page. The translation may involve dividing the content locator into segment tokens in a first language, followed by translating, transliterating or not changing a segment token. The processed tokens are then reassembled in a second language. The translation may be provided by a translation module through a content page provided by a network browser.

Abstract: A system and method for capitalizing translated text is provided. A capitalized source text is automatically translated to a target text. The target text is capitalized according to information in the capitalized source text.

Abstract: Systems and methods for enhancing machine translation post edit review processes are provided herein. According to some embodiments, methods for displaying confidence estimations for machine translated segments of a source document may include executing instructions stored in memory, the instructions being executed by a processor to calculate a confidence estimation for a machine translated segment of a source document, compare the confidence estimation for the machine translated segment to one or more benchmark values, associate the machine translated segment with a color based upon the confidence estimation for the machine translated segment relative to the one or more benchmark values, and provide the machine translated segment having the color in a graphical format, to a client device.

Abstract: Systems and methods for generating trust scores for translations are described herein. According to some embodiments, methods for generating a trust score for a translation may include establishing a trust score for at least a portion of a first translation of a source text translated by a trusted translation system, the trust score representing an accuracy level for the first translation, comparing the first translation of the source text generated by the trusted translation system to a second translation of the source text generated by an untrusted translation system, and determining a trust score for the second translation based upon the comparison.

Abstract: Embodiments of the present invention provide a system and method for providing a translation service. The method comprises providing a translation interface accessible via a network. The translation interface receives specialized data associated with a domain from a member. A text string written in a source language is received from the member via the translation interface. A domain-based translation engine is selected. The domain-based translation engine may be associated with a source language, a target language, and a domain. The text string is translated into the target language using, at least in part, the selected domain-based translation engine. The translated text string is transmitted to the member via the Internet. In some embodiments, a translation memory is generated based on the specialized data.

Abstract: Systems and methods for automatically modifying an annotated bilingual segment pair are provided. An annotated bilingual segment pair (“Pair”) may be modified to generate improved translation rules used in machine translation of documents from a source language to a target language. Because a single Pair may be used to translate a phrase, many Pairs are used in a machine translation system and manual correction of each model is impractical. Each Pair may be modified by re-labeling syntactic categories within the Pair, re-structuring a tree within the Pair, and/or re-aligning source words to target words within the Pair. In exemplary embodiments, many alternate Pairs (or portions thereof) are generated automatically, rule sequences corresponding to each are derived, and one or more rule sequences are selected. Using the selected rule sequence, a modified Pair is distilled.

Abstract: A method for tuning translation parameters in statistical machine translation based on ranking of the translation parameters is disclosed. According to one embodiment, the method includes sampling pairs of candidate translation units from a set of candidate translation units corresponding to a source unit, each candidate translation unit corresponding to numeric values assigned to one or more features, receiving an initial weighting value for each feature, comparing the pairs of candidate translation units to produce binary results, and using the binary results to adjust the initial weighting values to produce modified weighting values.

Abstract: Customers having a translation project to select a translation method from a variety of options, ranging from a completely human translation to a completely automated translation. For human translations, translation job information may be communicated through one or more network service modules which execute within a network service application, such as a web-based networking application. A network service module may register a user having an account with the network service application as a translator and communicate translation jobs to the user. One or more users who express interest in performing the translation are selected to perform a translation job, each job comprising at least a portion of the translation project. After a user provides a translation for the translation job, the translation is analyzed to generate a trust level prediction for the translation. A user translation profile may be updated after each translation to reflect the user's performance.

Abstract: A system, method, and computer program for generating and exploiting an approximate language model are provided. The method comprises generating a language model according to an approximate hashing technique. The language model comprises a plurality of event sequences in a target language, and each member of the plurality of the event sequences is associated with at least one count. The language model is queried for a member of the plurality of event sequences. A probability associated with the member of the plurality of event sequences is determined based on results of the query.

Abstract: A method, system and computer readable medium for hosting a community in multiple languages is disclosed. The method comprises receiving a request from a first member of a community to post content within the community. The content is written in a first language. Next, the content is translated into a second language. Login information is received from a second member. The second member may be associated with a second language. Lastly, the translated content is displayed to the second member. The method may further comprise receiving feedback from the second member relating to the quality of the translation or post edited versions of the automatically produced translation.

Abstract: A quality-prediction engine predicts a trust level associated with translational accuracy of a machine-generated translation. Training a quality-prediction may include translating a document in a source language to a target language by executing a machine-translation engine stored in memory to obtain a machine-generated translation. The training may further include comparing the machine-generated translation with a human-generated translation of the document. The human-generated translation is in the target language. Additionally, the training may include generating a mapping between features of the machine-generated translation and features of the human-generated translation based on the comparison.

Abstract: A special notation that extends the notion of IDL by weighted operators. The Weighted IDL or WIDL can be intersected with a language model, for example an n-gram language model or a syntax-based language model. The intersection is carried out by converting the IDL to a graph, and unfolding the graph in a way which maximizes its compactness.

Abstract: A training system for text to text application. The training system finds groups of documents, and identifies automatically similar documents in the groups which are similar. The automatically identified documents can then be used for training of the text to text application. The comparison uses reduced size versions of the documents in order to minimize the amount of processing.

Abstract: A statistical machine translation (MT) system may include a compound splitting module to split compounded words for more accurate translation. The compound splitting module select a best split for translation by the MT system.

Abstract: Tree transducers can be trained for use in probabilistic operations such as those involved in statistical based language processing. Given sample input/output pairs as training, and given a set of tree transducer rules, the information is combined to yield locally optimal weights for those rules. This combination is carried out by building a weighted derivation forest for each input/output pair and applying counting methods to those forests.