Binary coding for improved semantic search

One or more non-transitory computer storage media having computer-executable instructions embodied thereon that, when executed, by one or more processors, cause the one or more processors to perform a method, the method comprising:

• receiving, at a search engine, a query for one or more items, the query includes a plurality of terms;
• generating a first vector that represents the query and a second vector that represents a search result candidate in semantic space based on a semantic similarity to one or more terms of the plurality of terms;
• processing the first vector and the second vector through a machine learning model;
• in response to the processing, converting the first vector into a first binary vector of binary values in a binary semantic space and converting the second vector into a second binary vector of binary values in the binary semantic space, wherein each term of the plurality of terms is converted into a binary value, and each binary value of the of the first and second binary vectors being either a zero or one;
• in response to the converting, determining whether the second binary vector is within a threshold distance of the first binary vector within the binary semantic space;
• and
• based at least in part on determining that the second binary vector is within the threshold distance of the first binary vector, causing display of the search result candidate that has at least one term having an identical meaning to at least another term of the plurality of terms, wherein the at least one term and the at least another term do not structurally match.

A system comprising:

• one or more processors;
• and
• one or more non-transitory computer storage media storing computer-useable instructions that, when used by the one or more processors, cause the one or more processors to perform a method, the method comprising:
• receiving, at a search engine, a query that includes a plurality of terms;
• generating a first vector that represents the query and a second vector that represents a search result candidate in semantic space based on a semantic similarity to one or more terms of the plurality of terms;
• processing the first vector and the second vector through a machine learning model;
• in response to the processing, converting the first vector that represents the query into a first binary vector of binary values in a binary semantic space and converting the second vector into a second binary vector of binary values in the binary semantic space, wherein each term of the plurality of terms is converted into a binary value, and each binary value of the first and second binary vectors being either a zero or one;
• in response to the converting, determining whether the second binary vector is within a threshold distance of the first binary vector within the binary semantic space;
• and
• based at least in part on the determining that the second binary vector is within the threshold distance of the first binary vector, causing display of the search result candidate that includes at least a first term having a same meaning as at least a second term of the query, wherein each of the first term and the second term has a different word structure.