Abstract: Techniques are disclosed for training and/or utilizing an automatic post-editing model in correcting translation error(s) introduced by a neural machine translation model. The automatic post-editing model can be trained using automatically generated training instances. A training instance is automatically generated by processing text in a first language using a neural machine translation model to generate text in a second language. The text in the second language is processed using a neural machine translation model to generate training text in the first language. A training instance can include the text in the first language as well as the training text in the first language.

Abstract: Techniques are disclosed for training and/or utilizing an alignments and language model (“ALM”) in automatically determining an ALM score corresponding with natural language text generated using a natural language generation model. The natural language text generated using the natural language generation model can be based on a set of structured data. Additionally or alternatively, the ALM can include a fluency model portion and a semantics model portion. The fluency model portion can be used in determining the fluency and/or grammar of the text. The semantics model portion be used in evaluating the content of the natural language text with respect to the content of the structured data.

Abstract: Techniques are disclosed for training and/or utilizing an automatic post-editing model in correcting translation error(s) introduced by a neural machine translation model. The automatic post-editing model can be trained using automatically generated training instances. A training instance is automatically generated by processing text in a first language using a neural machine translation model to generate text in a second language. The text in the second language is processed using a neural machine translation model to generate training text in the first language. A training instance can include the text in the first language as well as the training text in the first language.

Abstract: Techniques are disclosed for training and/or utilizing an automatic post-editing model in correcting translation error(s) introduced by a neural machine translation model. The automatic post-editing model can be trained using automatically generated training instances. A training instance is automatically generated by processing text in a first language using a neural machine translation model to generate text in a second language. The text in the second language is processed using a neural machine translation model to generate training text in the first language. A training instance can include the text in the first language as well as the training text in the first language.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for providing a natural language based program interface to software applications. One of the methods includes, obtaining, via a natural language front end, a natural language query or a natural language update statement issued by a software application; converting the natural language query or natural language update statement into structured operations to be performed on APIs of a knowledge base; performing the structured operations on the APIs to produce a natural language output statement; and providing, via a natural language output interface, the natural language output statement to the software application. The knowledge base stores entity information according to a data schema and has structured APIs for use by software applications to query the knowledge base; the software applications are limited to communicating with the knowledge base through the interfaces provided by the natural language front end.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for processing word sequences using neural networks. One of the methods includes receiving a first sequence of words arranged according to a first order; and for each word in the first sequence, beginning with a first word in the first order: determining a topic vector that is associated with the word; generating a combined input from the word and the topic vector, and processing the combined input through one or more sequence modeling layers to generate a sequence modeling output for the word; and processing one or more of the sequence modeling outputs through an output layer to generate a neural network output for the first sequence of words.

Abstract: A server accesses an initial query associated with a classification, the classification corresponding to a likely intent of the initial query. The server obtains a set of queries, wherein each query in the set of queries is identified as having resulted in one or more users selecting a resource that was selected by one or more users in response to submitting the initial query. The server then determines a metric for one or more queries in the set of queries, wherein the metric for each of the one or more queries in the set of queries is based on a similarity between the respective query and the initial query. Next, the server selects a subset of queries from the set of queries based on the metric for each selected query satisfying a threshold and associates the selected subset of queries with the classification of the initial query.

Abstract: In some implementations, data that indicates multiple candidate transcriptions for an utterance is received. For each of the candidate transcriptions, data relating to use of the candidate transcription as a search query is received, a score that is based on the received data is provided to a trained classifier, and a classifier output for the candidate transcription is received. One or more of the candidate transcriptions may be selected based on the classifier outputs.

Abstract: Methods and systems for presenting content such as articles based on utility are provided. In one embodiment, a plurality of articles are determined, each article in the plurality of articles including article content and a corresponding preview icon, the preview icon defining a link to the corresponding article content when presented. For each article in the plurality of articles, a user experience utility value is determined. And for each article in the plurality of articles, an economic utility value is also determined. A ranked order of the articles is determined based upon each article's user experience utility value and economic utility value.

Abstract: In some implementations, data that indicates multiple candidate transcriptions for an utterance is received. For each of the candidate transcriptions, data relating to use of the candidate transcription as a search query is received, a score that is based on the received data is provided to a trained classifier, and a classifier output for the candidate transcription is received. One or more of the candidate transcriptions may be selected based on the classifier outputs.

Abstract: A computer-implemented method for constructing a profile for a target user is disclosed. The method comprises monitoring electronic communications across a network to identify at least one electronic communication that identifies a target user as an intended message recipient, extracting metadata from content associated with the at least one electronic communication; and constructing a user profile for the target user on the basis of the extracted metadata. It is assumed that the message senders send their messages, including associated content, to a given target user with the belief that the content is of some interest to the target user on the basis of some knowledge regarding the target user's personality, preferences, tastes and the like. In this manner, a profiling entity may indirectly construct a profile of the target user based on the content sent by one or more message senders to the target user.

Abstract: A computer-implemented method for constructing a profile for a target user is disclosed. The method comprises monitoring electronic communications across a network to identify at least one electronic communication that identifies a target user as an intended message recipient, extracting metadata from content associated with the at least one electronic communication; and constructing a user profile for the target user on the basis of the extracted metadata. It is assumed that the message senders send their messages, including associated content, to a given target user with the belief that the content is of some interest to the target user on the basis of some knowledge regarding the target user's personality, preferences, tastes and the like. In this manner, a profiling entity may indirectly construct a profile of the target user based on the content sent by one or more message senders to the target user.

Abstract: A method and apparatus for customizing content presented to individual users or user segments is provided. There may be three components, a web portal and toolbar component, a modeling component, and a scoring component. The web portal and toolbar component presents content items and collects data. The web portal and toolbar component generates user event data based on the user actions. The user event data is forwarded to the modeling component. The modeling component generates content scoring functions based on user event data and attributes of content items. Content scoring functions may be unique to individual user segments. The content scoring functions based on content features generate probability a content item will be viewed. The scoring component decides which content items are placed in a portal. The scoring component uses the scoring functions generated by the modeling component to rank content items in real time.

Abstract: A method and apparatus for customizing content presented to individual users or user segments is provided. There may be three components, a web portal and toolbar component, a modeling component, and a scoring component. The web portal and toolbar component presents content items and collects data. The web portal and toolbar component generates user event data based on the user actions. The user event data is forwarded to the modeling component. The modeling component generates content scoring functions based on user event data and attributes of content items. Content scoring functions may be unique to individual user segments. The content scoring functions based on content features generate probability a content item will be viewed. The scoring component decides which content items are placed in a portal. The scoring component uses the scoring functions generated by the modeling component to rank content items in real time.

Abstract: A computer-implemented method for constructing a profile for a target user is disclosed. The method comprises monitoring electronic communications across a network to identify at least one electronic communication that identifies a target user as an intended message recipient, extracting metadata from content associated with the at least one electronic communication; and constructing a user profile for the target user on the basis of the extracted metadata. It is assumed that the message senders send their messages, including associated content, to a given target user with the belief that the content is of some interest to the target user on the basis of some knowledge regarding the target user's personality, preferences, tastes and the like. In this manner, a profiling entity may indirectly construct a profile of the target user based on the content sent by one or more message senders to the target user.