摘要:
Query-correction pairs can be extracted from search log data. Each query-correction pair can include an original query and a follow-up query, where the follow-up query meets one or more criteria for being identified as a correction of the original query, such as an indication of user input indicating the follow-up query is a correction for the original query. The query-correction pairs can be segmented to identify bi-phrases in the query-correction pairs. Probabilities of corrections between the bi-phrases can be estimated based on frequencies of matches in the query-correction pairs. Identifications of the bi-phrases and representations of the probabilities of those bi-phrases can be stored in a probabilistic model data structure.
摘要:
Systems and methods for identifying translation pairs from web pages are provided. One disclosed method includes receiving monolingual web page data of a source language, and processing the web page data by detecting the occurrence of a predefined pattern in the web page data, and extracting a plurality of translation pair candidates. Each of the translation pair candidates may include a source language string and target language string. The method may further include determining whether each translation pair candidate is a valid transliteration. The method may also include, for each translation pair that is determined not to be a valid transliteration, determining whether each translation pair candidate is a valid translation. The method may further include adding each translation pair that is determined to be a valid translation or transliteration to a dictionary.
摘要:
Described is a technology for using clickthrough data (e.g., based on data of a query log) in learning a ranking model that may be used in online ranking of search results. Clickthrough data, which is typically sparse (because many documents are often not clicked or rarely clicked), is processed/smoothed into smoothed clickthrough streams. The processing includes determining similar queries for a document with incomplete (insufficient) clickthrough data to provide expanded clickthrough data for that document, and/or by estimating at least one clickthrough feature for a document when that document has missing (e.g., no) clickthrough data. Similar queries may be determined by random walk clustering and/or session-based query analysis. Features extracted from the clickthrough streams may be used to provide a ranking model which may then be used in online ranking of documents that are located with respect to a query.
摘要:
Candidate suggestions for correcting misspelled query terms input into a search application are automatically generated. A score for each candidate suggestion can be generated using a first decoding pass and paths through the suggestions can be ranked in a second decoding pass. Candidate suggestions can be generated based on typographical errors, phonetic mistakes and/or compounding mistakes. Furthermore, a ranking model can be developed to rank candidate suggestions to be presented to a user.
摘要:
A context modeling architecture that includes a context representation portion, which adapted to represent context as features, is provided. The features are specifiable at runtime of an application including the context representation portion.
摘要:
A probability distribution for a reference summary of a document is determined. The probability distribution for the reference summary is then used to generate a score for a machine-generated summary of the document.
摘要:
A search result ranker may be trained with automatically-generated samples. In an example embodiment, user interests are inferred from user interactions with search results for a particular query so as to determine respective relevance scores associated with respective query-identifier pairs of the search results. Query-identifier-relevance score triplets are formulated from the respective relevance scores associated with the respective query-identifier pairs. The query-identifier-relevance score triplets are submitted as training samples to a search result ranker. The search result ranker is trained as a learning machine with multiple training samples of the query-identifier-relevance score triplets.
摘要:
Systems and methods for selecting a ranker for statistical natural language processing are provided. One disclosed system includes a computer program configured to be executed on a computing device, the computer program comprising a data store including reference performance data for a plurality of candidate rankers, the reference performance data being calculated based on a processing of test data by each of the plurality of candidate rankers. The system may further include a ranker selector configured to receive a statistical natural language processing task and a performance target, and determine a selected ranker from the plurality of candidate rankers based on the statistical natural language processing task, the performance target, and the reference performance data.
摘要:
An algorithm that employs modified methods developed for optimizing differential functions but which can also handle the special non-differentiabilities that occur with the L1-regularization. The algorithm is a modification of the L-BFGS (limited-memory Broyden-Fletcher-Goldfarb-Shanno) quasi-Newton algorithm, but which can now handle the discontinuity of the gradient using a procedure that chooses a search direction at each iteration and modifies the line search procedure. The algorithm includes an iterative optimization procedure where each iteration approximately minimizes the objective over a constrained region of the space on which the objective is differentiable (in the case of L1-regularization, a given orthant), models the second-order behavior of the objective by considering the loss component alone, using a “line-search” at each iteration that projects search points back onto the chosen orthant, and determines when to stop the line search.
摘要:
A method of creating an index of web queries is discussed. The method includes receiving a first query representative of one or more symbolic characters and assigning the first query to a first data structure. A first text string representative of the first query is created and assigned to a second data structure. The first and second data structures are stored on a tangible computer readable medium.