Back to results
US 10,380,127 B2Granted
Candidate search result generation
AssigneeMicrosoft Technology Licensing, LLC
InventorsGanesh Venkataraman, Dhruv Arya, Aman Grover, Liang Zhang
Life of this patent
Estimating the term…
Abstract
A trained search system can be configured to retrieve a candidate subset of results, where the trained search system uses data extracted from a machine learning scheme. The machine learning scheme can be trained to identify results that are ranked by a computationally expensive algorithm, such as a ranking algorithm. When a query is received, the trained search system can be used to retrieve results instead of applying the computationally expensive ranking algorithm.
Claims20
A method comprising:
- generating, using one or more hardware processors, a ranked dataset by ranking a plurality of documents using a ranking scheme;
- identifying which of the ranked plurality of documents have associated tracked positive user actions within a graphical user interface;
- dividing the ranked plurality of documents into a top-ranked portion and a bottom-ranked position;
- using a machine learning engine to train a decision tree to classify documents as either included or excluded from the top-ranked portion of the ranked documents, the decision tree weighted so that documents which receive further tracked positive user actions are more likely to be included in the top-ranked portion;
- extracting weighting information from the trained decision tree;
- training a search engine to weight one or more input data fields with weights from the trained machine learning scheme, the one or more input data fields include one or more query fields;
- receiving a query for processing against the plurality of documents;
- generating, using the trained search engine, a candidate subset of the plurality of documents as results for the query in lieu of using the ranking scheme for results of the query;
- and
- providing one or more documents from the candidate subset to the client device.
7 more elements
A system comprising:
- one or more hardware processors of a machine;
- and
- a memory storing instructions that, when executed by the one or more hardware processors, cause the machine to perform operations comprising:
- generating, using one or more hardware processors, a ranked dataset by ranking a plurality of documents using a ranking scheme;
- identifying which of the ranked plurality of documents have associated tracked positive user actions within a graphical user interface;
- dividing the ranked plurality of documents into a top-ranked portion and a bottom-ranked position;
- using a machine learning engine to train a decision tree to classify documents as either included or excluded from the top-ranked portion of the ranked documents, the decision tree weighted so that documents which receive further tracked positive user actions are more likely to be included in the top-ranked portion;
- extracting weighting information from the trained decision tree;
- training a search engine to weight one or more input data fields with weights from the trained machine learning scheme, the one or more input data fields include one or more query fields;
- receiving a query for processing against the plurality of documents;
- generating, using the trained search engine, a candidate subset of the plurality of documents as results for the query in lieu of using the ranking scheme for results of the query;
- and
- providing one or more documents from the candidate subset to the client device.
10 more elements
A machine-readable storage medium embodying instructions that, when executed by a machine, cause the machine to perform operations comprising:
- generating, using one or more hardware processors, a ranked dataset by ranking a plurality of documents using a ranking scheme;
- identifying which of the ranked plurality of documents have associated tracked positive user actions within a graphical user interface;
- dividing the ranked plurality of documents into a top-ranked portion and a bottom-ranked position;
- using a machine learning engine to train a decision tree to classify documents as either included or excluded from the top-ranked portion of the ranked documents, the decision tree weighted so that documents which receive further tracked positive user actions are more likely to be included in the top-ranked portion;
- extracting weighting information from the trained decision tree;
- training a search engine to weight one or more input data fields with weights from the trained machine learning scheme, the one or more input data fields include one or more query fields;
- receiving a query for processing against the plurality of documents;
- generating, using the trained search engine, a candidate subset of the plurality of documents as results for the query in lieu of using the ranking scheme for results of the query;
- and
- providing one or more documents from the candidate subset to the client device.
7 more elements
Dependencies are inferred from the claim text.