Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a machine learning model having a set of local model parameters and a set of global model parameters under a partially local federated learning framework. One of the methods include maintaining local data and data defining the local model parameters; receiving data defining current values of the global model parameters; determining, based on the local data, the local model parameters, and the current values of the global model parameters, current values of the local model parameters; determining, based on the local data, the current values of the local model parameters, and the current values of the global model parameters, updated values of the global model parameters; generating, based on the updated values of the global model parameters, parameter update data defining an update to the global model parameters; and transmitting the parameter update data.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for evaluating substitute terms. One of the methods includes receiving a query having an original term and determining one or more substitution contexts for the original term, wherein a substitution context includes one or more context terms and an indication of a position in the query of the original term and the one or more context terms. The substitution contexts are classified into a first category or a second category based on a respective score of each substitution context. The original term is associated with one or more substitution contexts in the first category.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training recognition canonical representations corresponding to named-entity phrases in a second natural language based on translating a set of allowable expressions with canonical representations from a first natural language, which may be generated by expanding a context-free grammar for the allowable expressions for the first natural language.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for evaluating substitute terms. One of the methods includes receiving a query having an original term and determining one or more substitution contexts for the original term, wherein a substitution context includes one or more context terms and an indication of a position in the query of the original term and the one or more context terms. The substitution contexts are classified into a first category or a second category based on a respective score of each substitution context. The original term is associated with one or more substitution contexts in the first category.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for evaluating substitute terms. One of the methods includes receiving a query and identifying an original term occurring in the query. One or more substitution contexts are determined for the original term, wherein a substitution context includes one or more context terms and an indication of a position in the query of the original term and the one or more context terms. A score is determined for each substitution context of the one or more substitution contexts. One or more substitution contexts are selected based on the score of each substitution context. One or more substitute terms for the original term in the context of the one or more selected substitution contexts are determined.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training recognition canonical representations corresponding to named-entity phrases in a second natural language based on translating a set of allowable expressions with canonical representations from a first natural language, which may be generated by expanding a context-free grammar for the allowable expressions for the first natural language.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for evaluating substitute terms. One of the methods includes receiving a query and identifying an original term occurring in the query. One or more substitution contexts are determined for the original term, wherein a substitution context includes one or more context terms and an indication of a position in the query of the original term and the one or more context terms. A score is determined for each substitution context of the one or more substitution contexts. One or more substitution contexts are selected based on the score of each substitution context. One or more substitute terms for the original term in the context of the one or more selected substitution contexts are determined.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for evaluating substitute terms. One of the methods includes receiving a query having an original term and determining one or more substitution contexts for the original term, wherein a substitution context includes one or more context terms and an indication of a position in the query of the original term and the one or more context terms. The substitution contexts are classified into a first category or a second category based on a respective score of each substitution context. The original term is associated with one or more substitution contexts in the first category.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for evaluating substitute terms. One of the methods includes selecting a first term and a second term. A first co-occurrence frequency is determined for co-occurring terms in search queries that include the first term. A first vector is generated for the first term using the first co-occurrence frequencies. A second co-occurrence frequency is determined for the co-occurring terms in the search queries that include the first term adjacent to the second term. A second vector is generated for the second term using the second co-occurrence frequencies. A score for the second term as a context for a substitution rule based on the first term is computed, wherein the score is based on a comparison between the first vector and the second vector.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for identifying a particular query term of an original search query, identifying a candidate synonym for the particular query term in context with an other non-adjacent query term of the original search query that is not adjacent to the particular query term in the original search query, accessing stored data that specifies, for a pair of terms that includes the particular query term and the candidate synonym of the particular query term, a respective confidence value for the other non-adjacent query term, determining that, in the stored data, the confidence value for the other non-adjacent query term satisfies a threshold, and determining to revise the original search query to include the candidate synonym of the particular query term, based on determining that the confidence value the other non-adjacent query term satisfies the threshold.