Methods, systems, and media for automated compliance determination of content items

Patent metadata

Field	Value
Publication number	US-2021182902-A1
Application number	US-201916710954-A
Country	US
Kind code	A1
Filing date	Dec 11, 2019
Priority date	Dec 11, 2019
Publication date	Jun 17, 2021
Grant date	—

Abstract

Methods, systems, and media for automated compliance determination of content items are provided. In some embodiments, the method comprises: receiving, at a server from a user device associated with a user, a request to provide a branded content item on a media content platform; in response to receiving the request, generating a transcript of a speech portion of the branded content item; generating a plurality of candidate word sequences based on the transcript; selecting a candidate word sequence from the plurality of candidate word sequences based on a similarity that is determined by comparing each of the plurality of candidate word sequences with each of a plurality of target word sequences; in response to selecting the candidate word sequence, applying a model to the selected candidate word sequence to determine whether the selected candidate word sequence contains a first disclosure statement in accordance with one or more disclosure requirements and applying the model to a content description associated with the branded content item to determine whether the content description contains a second disclosure statement in accordance with the one or more disclosure requirements; and associating the branded content item with a compliance indicator that indicates the branded content item is compliant with the one or more disclosure requirements in response to the model indicating that the selected candidate word sequence contains the first disclosure statement and in response to the model indicating that the content description contains the second disclosure statement.

First claim

What is claimed is:

1 . A computer-implemented method for compliance checking content items, the method comprising:
receiving, at a server from a user device associated with a user, a request to provide a branded content item on a media content platform; in response to receiving the request, generating a transcript of a speech portion of the branded content item; generating a plurality of candidate word sequences based on the transcript; selecting a candidate word sequence from the plurality of candidate word sequences based on a similarity that is determined by comparing each of the plurality of candidate word sequences with each of a plurality of target word sequences; in response to selecting the candidate word sequence, applying a model to the selected candidate word sequence to determine whether the selected candidate word sequence contains a first disclosure statement in accordance with one or more disclosure requirements and applying the model to a content description associated with the branded content item to determine whether the content description contains a second disclosure statement in accordance with the one or more disclosure requirements; and associating the branded content item with a compliance indicator that indicates the branded content item is compliant with the one or more disclosure requirements in response to the model indicating that the selected candidate word sequence contains the first disclosure statement and in response to the model indicating that the content description contains the second disclosure statement.

2 . The computer-implemented method of claim 1 , wherein the transcript of the speech portion of the branded content item is generated by transmitting a content identifier of the branded content item to a speech-to-text converter that converts speech in the branded content item to text in the transcript.

3 . The computer-implemented method of claim 1 , wherein a time period of the speech portion is selected based on the one or more disclosure requirements requiring that the first disclosure statement is spoken within the time period of the branded content item.

4 . The computer-implemented method of claim 1 , wherein each of the plurality of candidate word sequences generated from the transcript is a particular length of words.

5 . The computer-implemented method of claim 1 , wherein the similarity is determined by:
generating a plurality of candidate vectors, wherein each of the plurality of candidate word sequences is embedded into a candidate vector; generating a plurality of target vectors, wherein each of the plurality of target word sequences is embedded into a target vector; and comparing each of the plurality of candidate vectors with each of the plurality of target vectors determine a similarity score, wherein the candidate word sequence having a highest similarity score is selected.

6 . The computer-implemented method of claim 5 , wherein the similarity score is determined by calculating cosine similarity between each of the plurality of candidate vectors and each of the plurality of target vectors.

7 . The computer-implemented method of claim 1 , wherein, in response to the model indicating that the selected candidate word sequence contains the first disclosure statement, the compliance indicator is modified to indicate that the branded content item is non-compliant with the one or more disclosure requirements.

8 . The computer-implemented method of claim 1 , wherein, in response to the model indicating that the content description contains the second disclosure statement, the compliance indicator is modified to indicate that the branded content item is non-compliant with the one or more disclosure requirements.

9 . The computer-implemented method of claim 1 , wherein, in response to determining that none of the plurality of candidate word sequences has a similarity with one of the plurality of target word sequences greater than the similarity threshold value, the compliance indicator is modified to indicate that the branded content item is non-compliant with the one or more disclosure requirements.

10 . The computer-implemented method of claim 1 , wherein the model is applied to the content description associated with the branded content item to determine whether the content description contains the second disclosure statement in accordance with the one or more disclosure requirements in response to an output of the model determining that the selected candidate word sequence is likely to contain the first disclosure statement in accordance with the one or more disclosure requirements.

11 . The computer-implemented method of claim 1 , wherein the compliance indicator is set to indicate that the branded content item is compliant with the one or more disclosure requirements in response to determining that an output of the model is greater than a threshold compliance value.

12 . The computer-implemented method of claim 1 , further comprising transmitting the branded content item and the compliance indicator to a reviewing user prior to providing the branded content item on the media content platform.

13 . The computer-implemented method of claim 12 , further comprising determining whether to transmit the branded content item and the compliance indicator to the reviewing user based on a risk tolerance associated with the branded content item.

14 . The computer-implemented method of claim 12 , further comprising determining whether to transmit the branded content item and the compliance indicator to the reviewing user based on a confidence value associated with the compliance indicator.

15 . The computer-implemented method of claim 1 , further comprising allowing the branded content item to be published on the media content platform based on the compliance indicator indicating that the branded content item is compliant with the one or more disclosure requirements.

16 . The computer-implemented method of claim 1 , further comprising inhibiting the branded content item from being published on the media content platform based on the compliance indicator indicating that the branded content item is non-compliant with the one or more disclosure requirements.

17 . The computer-implemented method of claim 1 , further comprising transmitting a notification to the user of the user device that recommends modifications to the branded content item based on the compliance indicator indicating that the branded content item is non-compliant with the one or more disclosure requirements.

18 . The computer-implemented method of claim 1 , further comprising determining that the speech portion of the branded content item is in a first language, wherein the speech portion is translated from the first language to a second language prior to generating the transcript of the speech portion of the branded content item.

19 . A system for compliance checking content items, the system comprising:
a memory; and a hardware processor that, when executing computer executable instructions stored in the memory, is configured to:
receive, at a server from a user device associated with a user, a request to provide a branded content item on a media content platform;
in response to receiving the request, generate a transcript of a speech portion of the branded content item;
generate a plurality of candidate word sequences based on the transcript;
select a candidate word sequence from the plurality of candidate word sequences based on a similarity that is determined by comparing each of the p

Assignees

• Google Llc

Inventors

• Scott-Green Henry
• De Ridder Michael
• Gaffney T J
• Mulford Brian
• Tveit Amund
• Delaite Antoine
• Sundar Preethi Puducheri

Classifications

• G06N20/00

  Machine learning · CPC title

• G06F40/20

  Natural language analysis (semantic analysis of natural language G06F40/30) · CPC title

• G10L15/26

  Speech to text systems (G10L15/08 takes precedence) · CPC title

• G06Q30/0248 Primary

  Avoiding fraud · CPC title

• G10L15/22

  Procedures used during a speech recognition process, e.g. man-machine dialogue · CPC title