Named entity disambiguation using entity distance in a knowledge graph

This invention helps computers understand which specific real-world thing a name or phrase in a text refers to, like distinguishing between "Apple" the company and "apple" the fruit. It works by first finding these names in the text, then looking up all possible meanings for each name in a large database of facts. It then compares these possible meanings based on their numerical representations to find the best match for each name, effectively linking the text to specific entries in the fact database. The claims specifically detail how these comparisons are made, either by averaging the numerical representations or by comparing them in pairs.