Search engine domain transfer
AssigneeInternational Business Machines Corporation
InventorsTin K. Ho, Luis A. Lastras-Montano, Vinith Misra
Life of this patent
Estimating the term…
Abstract
A mechanism is provided for search engine domain transfer. The mechanism receives an input query to search a specialized domain from a user and performs a general-domain search based on the input query to generate a set of general-domain results. The mechanism generates a feature vector based on the general-domain results and generates a score for each document within the specialized domain based on the feature vector. The mechanism generates a ranked result set of documents based on the scores of the documents in the specialized domain and presents the ranked result set to the user.
Claims12
A computer program product comprising a computer readable storage medium having a computer readable program stored therein, wherein the computer readable program, when executed on at least one processor of a data processing system, causes the at least one processor to implement a specialized-domain search engine with domain transfer, wherein the computer readable program causes the at least one processor to:
- receive, by the specialized-domain search engine executing on the at least one processor of the data processing system, an input query to search a narrow domain corpus of documents from a user;
- perform, by the specialized-domain search engine, a general-domain search based on the input query to generate a set of general-domain results, wherein performing the general-domain search comprises:
- transforming the input query into a plurality of general-domain queries, wherein transforming the input query further comprises splitting the input query into an original query version and a keyword-only version and concatenating each of a plurality of context strings to the original query version and to the keyword-only version, wherein the plurality of context strings are relevant to the narrow domain corpus of documents;
- sending the plurality of general-domain queries to a plurality of instances of a general-domain search engine in parallel, wherein the plurality of instances of the general-domain search engine search a general domain corpus of documents based on the plurality of general-domain queries;
- and receiving the set of general-domain results from the plurality of instances of the general-domain search engine with respect to a general domain corpus of documents;
- generate, by a featurizer component executing within the specialized-domain search engine, a feature vector based on the general-domain results, wherein generating the feature vector comprises generating a sub-vector for each of the plurality of general-domain queries and concatenating sub-vectors for the plurality of general-domain queries to form the feature vector, wherein each sub-vector comprises the input query with a different set of features masked off;
- generate, by a classifier component executing within the specialized-domain search engine, a score for each document within the narrow domain corpus of documents by comparing the feature vector to each document within the narrow domain corpus of documents;
- generate, by the specialized-domain search engine, a ranked result set of documents from the narrow domain corpus of documents based on the scores of the documents in the narrow domain corpus of documents;
- and
- present, by the specialized-domain search engine, the ranked result set of documents to the user.
An apparatus comprising:
- a processor;
- and
- a memory coupled to the processor, wherein the memory comprises instructions which, when executed by the processor, cause the processor to implement a specialized-domain search engine with domain transfer, wherein the instructions cause the at least one processor to:
- receive, by the specialized-domain search engine executing on the at least one processor of the data processing system, an input query to search a narrow domain corpus of documents from a user;
- perform, by the specialized-domain search engine, a general-domain search based on the input query to generate a set of general-domain results, wherein performing the general-domain search comprises:
- transforming the input query into a plurality of general-domain queries, wherein transforming the input query further comprises splitting the input query into an original query version and a keyword-only version and concatenating each of a plurality of context strings to the original query version and to the keyword-only version, wherein the plurality of context strings are relevant to the narrow domain corpus of documents;
- sending the plurality of general-domain queries to a plurality of instances of a general-domain search engine in parallel, wherein the plurality of instances of the general-domain search engine search a general domain corpus of documents based on the plurality of general-domain queries;
- and receiving the set of general-domain results from the plurality of instances of the general-domain search engine with respect to a general corpus of documents;
- generate, by a featurizer component executing within the specialized-domain search engine, a feature vector based on the general-domain results, wherein generating the feature vector comprises generating a sub-vector for each of the plurality of general-domain queries and concatenating sub-vectors for the plurality of general-domain queries to form the feature vector, wherein each sub-vector comprises the input query with a different set of features masked off;
- generate, by a classifier component executing within the specialized-domain search engine, a score for each document within the narrow domain corpus of documents by comparing the feature vector to each document within the narrow domain corpus of documents;
- generate, by the specialized-domain search engine, a ranked result set of documents from the narrow domain corpus of documents based on the scores of the documents in the narrow domain corpus of documents;
- and
- present, by the specialized-domain search engine, the ranked result set of documents to the user.
Dependencies are inferred from the claim text.