Preliminary ranker for scoring matching documents
AssigneeMicrosoft Technology Licensing, LLC
InventorsMichael Joseph Hopcroft, Robert Lovejoy Goodwin, Andrija Antonijevic
Life of this patent
Estimating the term…
Abstract
The technology described herein provides for preliminary ranking of matching documents for a search query. A preliminary ranker uses score tables for scoring each matching document based on its relevant to a search query. The score table for a document stores pre-computed data used to derive a frequency of terms and other information in the document. The preliminary ranker uses the score table for each matching document and the terms form the search query to determine a score for each matching document. The lowest scoring documents are removed from further consideration by a final ranker.
Claims20
A computer-implemented method for scoring a plurality of documents based on relevancy to a search query, the method comprising:
- accessing a table that is associated with a document found to be relevant to at least a portion of the search query, wherein the table stores pre-computed data used to derive a frequency of each term of a subset of terms in the document, and wherein each term of the subset of terms occurs in the document more than once;
- algorithmically determining the frequency of at least one term that corresponds to the search query;
- and
- based at least on the frequency and other data associated with the document and the search query, computing a score of the document in relation to the at least one term that corresponds to the search query.
A computer-implemented method comprising:
- accessing a table having a plurality of slots that store data associated with a document;
- for a first slot of the table, comparing a portion of a first hash key associated with a first term whose corresponding data is stored in the first slot with a portion of a second hash key associated with a second term whose data is being considered to be added to the first slot;
- if the portion of the first hash key matches the portion of the second hash key,
- (1) determining, from the data in the table, that a frequency of the second term in the document is greater than the frequency of the first term in the document, and
- (2) storing data associated with the frequency of the second term in the first slot of the table;
- and
- if the portion of the first hash key does not match the portion of the second hash key, not storing data corresponding to the second term in the first slot of the table.
One or more computer storage media storing computer-useable instructions that, when used by one or more computing devices, cause the one or more computing devices to perform a method for scoring a plurality of documents based on relevancy to a search query, the method comprising:
- accessing a table that stores data corresponding to a document, the data used to derive one or more pre-computed components that contribute to a score of the document in relation to the search query, the one or more pre-computed components comprising:
- (1) a frequency of one or more terms in the document, wherein the one or more terms are found in the document at least two times, and (2) at least one portion of the document in which the one or more terms are located;
- computing scores for each of the one or more pre-computed components including the frequency of the one or more terms in the document and the at least one portion of the document in which the one or more terms are located;
- computing scores for each of one or more real-time components that are computed after the search query is entered;
- and
- computing a final score for the document in relation to the search query based on the scores for the one or more pre-computed components and the one or more real-time components.
Dependencies are inferred from the claim text.