| CPC G06F 40/289 (2020.01) [G06F 17/18 (2013.01); G06N 20/00 (2019.01)] | 18 Claims |
|
1. A method comprising:
receiving, in a machine learning model, a corpus comprising a plurality of words comprising natural language terms, wherein the machine learning model comprises a plurality of layers configured to extract a plurality of keywords out of the corpus and further comprises a retrospective layer;
identifying, in the plurality of layers, a first keyword from the corpus and a second keyword from the corpus;
assigning the first keyword a first probability and the second keyword a second probability, wherein each probability is a corresponding likelihood that a corresponding keyword is to be included in a key phrase;
determining, in the retrospective layer, a first probability modifier that modifies the first probability based on a first dependence relationship between the second keyword being placed after the first keyword,
wherein the first probability modifier punishes the first keyword when the first keyword depends on the second keyword and the second keyword comes after the first keyword in the corpus, and
wherein punishes comprises reducing a probability that the first keyword is in the key phrase;
modifying the first probability using the first probability modifier to form a first modified probability;
using the first modified probability to determine whether the first keyword and the second keyword together form the key phrase; and
storing the key phrase in a non-transitory computer readable storage medium.
|