SYSTEMS AND METHODS FOR WORD SEGMENTATION BASED ON A COMPETING NEURAL CHARACTER LANGUAGE MODEL

摘要

Execute a string algorithm on a title associated with a product to identify at least one product type associated with the product, use a machine learning algorithm to predict at least one product type associated with a product based on the title, during identification or prediction A system and method are provided for detecting inaccuracies in a product title, comprising detecting inaccuracies in a title based on at least one, and outputting a message to a remote device indicating that the title includes inaccuracies. Executing the string algorithm includes receiving a set of strings, generating a tree based on the received set of strings, receiving a title, and traversing the generated tree using the title to find a match. can do. Using a machine learning algorithm may include identifying words in a title, learning a vector representation for each character n-gram of each word, and summing each character n-gram.