Synonym generation

A computer-implemented method for generating synonym pairs, comprising:

• generating a first vector representation and a second vector representation of a first phrase and a second phrase, respectively, wherein the first vector representation of the first phrase incorporates a probability distribution of user actions made in connection with the first phrase and the second vector representation of the second phrase incorporates a probability distribution of user actions made in connection with the second phrase;
• calculating a similarity vector comprising a similarity score that represents a similarity between the first vector representation and the second vector representation, the similarity vector being an n-dimensional vector of real numbers that represents a multi-dimensional similarity between the first phrase and the second phrase;
• determining whether the first phrase and the second phrase are a synonym pair by comparing one or more components of the similarity vector against a number threshold;
• and
• if the first phrase and the second phrase constitute a synonym pair, reporting that the first phrase and the second phrase constitute the synonym pair.

A computer system for generating synonym pairs comprising:

• a synonym generation computer device comprising a memory and a processing device, the memory storing computer-readable instructions directing the processing device to:
• retrieve one or more data sources containing at least a first phrase and a second phrase;
• generate a first vector representation of the first phrase, wherein the first vector representation of the first phrase incorporates a probability distribution of user actions made in connection with the first phrase;
• generate a second vector representation of the second phrase, wherein the second vector representation of the second phrase incorporates a probability distribution of user actions made in connection with the second phrase;
• determine at least one similarity score for the second vector representation with respect to the first vector representation;
• generate a similarity vector comprising the at least one similarity score by generating an n-dimensional vector of real numbers that represents a multi-dimensional similarity between the first phrase and the second phrase;
• filter the first phrase and the second phrase by comparing one or more components of the similarity vector against a number threshold;
• and
• report whether the first phrase and the second phrase constitute a synonym pair.