Abstract: Techniques are provided to allow users of a social network to have multilingual profiles (or profiles in second languages that are different than the users' native, or original, profile languages). In one technique, a translation model is applied to one or more data items (in a first language) in a user's profile to generate translated data items in a second language. The translated data items are displayed to the user (or an online social “friend” of the user) and the user is allowed to select one of the translated data items to include in the user's “second language” profile. The selection may then be used to improve the translation model.

Abstract: Techniques are provided to allow users of a social network to have multilingual profiles (or profiles in second languages that are different than the users' native, or original, profile languages). In one technique, a translation model is applied to one or more data items (in a first language) in a user's profile to generate translated data items in a second language. The translated data items are displayed to the user (or an online social “friend” of the user) and the user is allowed to select one of the translated data items to include in the user's “second language” profile. The selection may then be used to improve the translation model.

Abstract: A system stores a table mapping users to attributes, and stores a second table mapping the users to products associated with a source domain. The system determines a set of top scoring products for each of the attributes, and creates, using the top scoring products, a model that is predictive of an activity in a target domain, the target domain being separate from the source domain. The system detects a behavior from a particular user accessing the target domain, and generates a personalized prediction for the particular user based on the model, in response to the detecting the behavior.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for retrieving documents. One of the methods includes receiving a search query that includes a first query term and an adjacent, second query term, and a substitute term for the first query term. A determination is made that the first query term and the substitute term satisfy one or more predetermined criteria and that a resource does not include the first query term. The resource is selected to be scored only if the substitute term occurs adjacent to the second term in the resource.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for identifying and scoring restricted-locality synonyms. In one aspect, a method includes receiving a search query including a query term and a synonym for the query term. The synonym is evaluated using one or more predetermined criteria and based on the evaluation is designated as a restricted-locality synonym. A first scoring model is selected that specifies how to score occurrences of restricted-locality synonyms in documents. A document is received that is identified as being responsive to the search query. A document score for the document is determined using the first scoring model.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for identifying and scoring restricted-locality synonyms. In one aspect, a method includes receiving a search query including a query term and a synonym for the query term. The synonym is evaluated using one or more predetermined criteria and based on the evaluation is designated as a restricted-locality synonym. A first scoring model is selected that specifies how to score occurrences of restricted-locality synonyms in documents. A document is received that is identified as being responsive to the search query. A document score for the document is determined using the first scoring model.

Abstract: A discriminatively trained word order model is used to identify a most likely word order from a set of word orders for target words translated from a source sentence. For each set of word orders, the discriminatively trained word order model uses features based on information in a source dependency tree and a target dependency tree and features based on the order of words in the word order. The discriminatively trained statistical model is trained by determining a translation metric for each of a set of N-best word orders for a set of target words. Each of the N-best word orders are projective with respect to a target dependency tree and the N-best word orders are selected using a combination of an n-gram language model and a local tree order model.

Abstract: A discriminatively trained word order model is used to identify a most likely word order from a set of word orders for target words translated from a source sentence. For each set of word orders, the discriminatively trained word order model uses features based on information in a source dependency tree and a target dependency tree and features based on the order of words in the word order. The discriminatively trained statistical model is trained by determining a translation metric for each of a set of N-best word orders for a set of target words. Each of the N-best word orders are projective with respect to a target dependency tree and the N-best word orders are selected using a combination of an n-gram language model and a local tree order model.