Utilizing regular expression embeddings for named entity recognition systems

Patent metadata

Field	Value
Publication number	US-2021056099-A1
Application number	US-202017022594-A
Country	US
Kind code	A1
Filing date	Sep 16, 2020
Priority date	Aug 23, 2019
Publication date	Feb 25, 2021
Grant date	—

Abstract

Various embodiments are directed to a system that utilizes regular expression (regex) to recognize at least portions of characters, words, text, numbers, etc. in a structured or unstructured dataset, any patterns associated therewith, and/or similarities between the determined patterns. In examples, a regex-based pattern recognition platform may receive a dataset and determine whether at least a first regex pattern and a second regex pattern can be identified. The occurrences of the first and second regex patterns and the frequency of those occurrences may reveal something about the dataset itself or any patterns contained therein.

First claim

What is claimed is:

1 . A method comprising:
determining, via one or more processors, whether a dataset contains a first regex pattern of a regex list; determining, via the one or more processors, a number of occurrences of the first regex pattern in the dataset; generating, via the one or more processors, a vector, the vector specifying a vector position and a regex value for at least the first regex pattern; incrementing, via the one or more processors, the regex value corresponding to the first regex pattern by a predefined value for each occurrence of the first regex pattern in the dataset; determining, via the one or more processors, any false matches between the first regex pattern of the regex list and the dataset; refining, based on any determined false matches, the regex list via a machine learning model or a classification model; and providing, via the one or more processors, the vector to at least one entity recognition system.

2 . The method of claim 1 , further comprising:
determining, via the one or more processors, whether the dataset contains a second regex pattern of the regex list; determining, via the one or more processors, a number of occurrences of the second regex pattern in the dataset; and incrementing, via the one or more processors, a regex value corresponding to the second regex pattern by the predefined value for each occurrence of the second regex pattern in the dataset.

3 . The method of claim 1 , further comprising performing a normalization on the vector.

4 . The method of claim 1 , further comprising identifying any personally identifiable information (PII) in the dataset and replacing the PII with a regex pattern from the regex list corresponding to the PII.

5 . The method of claim 1 , wherein the predefined value associated with the incrementing is one.

6 . The method of claim 3 , further comprising providing the normalized vector to a machine learning model and/or a classification model, and wherein the vector is used to train the machine learning model and/or the classification model.

7 . The method of claim 1 , wherein the at least one named entity recognition system determines whether (i) one or more portions of the received dataset includes personally identifiable information (PII) or (ii) classifies the one or more portions of the received dataset as PII.

8 . The method of claim 1 , wherein the first regex defines a personally identifiable information (PII) pattern.

9 . A method comprising:
determining, via one or more processors, whether a dataset comprises unstructured text; determining, via the one or more processors, that at least a portion of the unstructured text corresponds to a regex pattern of a regex list; replacing, via the one or more processors, the portion of the unstructured text with an encoding associated with the regex pattern to generate a modified dataset; and providing at least the modified dataset to at least one entity recognition system.

10 . The method of claim 9 , wherein the portion of the unstructured text is a first character, the first character being an alphanumeric character or special character.

11 . The method of claim 10 , wherein the encoding is a second character, the second character being an alphanumeric character or a special character.

12 . The method of claim 9 , wherein the portion of the unstructured text is a string of two or more consecutive characters, the two or more consecutive characters of the strings being alphanumeric characters or special characters.

13 . The method of claim 12 , wherein the encoding is a word of at least one character, the at least one character of the word being an alphanumeric character or a special character.

14 . The method of claim 9 , further comprising:
determining, via the one or more processors, any false matches between the regex pattern of the regex list and the dataset; refining, based on any determined false matches, the regex list via a machine learning model or a classification model.

15 . At least one non-transitory computer-readable storage medium storing computer-readable program code executable by at least one processor to:
determine whether a dataset comprises unstructured text; determine that at least a portion of the unstructured text corresponds to a regex pattern of a regex list; replace the portion of the unstructured text with an encoding associated with the regex pattern to generate a modified dataset; and provide at least the modified dataset to at least one entity recognition system.

16 . The at least one non-transitory computer-readable storage medium of claim 15 , wherein the portion of the unstructured text is a first character, the first character being an alphanumeric character or special character.

17 . The at least one non-transitory computer-readable storage medium of claim 16 , wherein the encoding is a second character, the second character being an alphanumeric character or a special character.

18 . The at least one non-transitory computer-readable storage medium of claim 15 , wherein the portion of the unstructured text is a string of two or more consecutive characters, the two or more consecutive characters of the strings being alphanumeric characters or special characters.

19 . The at least one non-transitory computer-readable storage medium of claim 18 , wherein the encoding is a word of at least one character, the at least one character of the word being an alphanumeric character or a special character.

20 . The at least one non-transitory computer-readable storage medium of claim 15 , wherein the computer-readable program code further causes the at least one processor to:
determine any false matches between the regex pattern of the regex list and the dataset; and refine, based on any determined false matches, the regex list via a machine learning model or a classification model.

Assignees

• Capital One Services Llc

Inventors

• Goodsitt Jeremy Edward
• Walters Austin Grant
• Farivar Reza
• Watson Mark Louis
• Truong Anh
• Rafferty Galen
• Pham Vincent

Classifications

• G06N20/20

  Ensemble learning · CPC title

• G06F16/35 Primary

  Clustering; Classification · CPC title

• G06F16/243 Primary

  Natural language query formulation · CPC title

• G06F16/313

  Selection or weighting of terms for indexing · CPC title

• G06F16/3329

  Natural language query formulation · CPC title