# US 2020/0402098 A1 · Systems and methods for assessing advertisement

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2020/0402098 A1
- Document ID: US20200402098A1
- Status: Published application
- Kind: A1
- Application number: 16/760,621
- Filed: 2019-09-16
- Published: 2020-12-24
- Earliest claimed: 2018-09-17
- Assignees: Google Inc.
- Applicants: Google LLC
- Inventors: Sylvanus Garnet Bent, III; Prahlad Fogla; Jamie Nicole Powell; Shu Niu; Nam Hoang Mai; Tristan Dennen; Sean Burroughs Johnston; Siva Kumar Gorantla; Suzanna Whiteside Shwert; Maxwell Schram; Weikun Liang
- Web page: https://patentgrep.com/patent/US20200402098A1

## CPC

- G06Q 30/0242: Determining effectiveness of advertisements
- G06Q 30/0256: {User search}
- G06F 16/41: Indexing; Data structures therefor; Storage structures
- G06F 40/284: Lexical analysis, e.g. tokenisation or collocates
- G06N 3/049: Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G06Q 30/0244: {Optimization}
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Systems and methods are disclosed for dynamically analyzing and providing the quality of one or more content items at the time, or substantially close to the time, they are received by a data processing system. The systems and methods described herein can maintain and update the quality score for improving previously created content items after they have been published. The one or more content items can include one or more assets (e.g., one or more headlines, one or more descriptions, images, video, etc.). The data processing system can use numerical analysis methods to determine an overall quality (e.g., estimated clicks) of the content items received by the data processing system using a trained model.

## Claims

### Claim 1

A method for assessing content items using numerical criteria comprising:
receiving, by an asset reception module, a plurality of text strings, each of the plurality of text strings comprising a token;
receiving, by the asset reception module, a plurality of target keywords;
retrieving, by an asset analysis module, one or more synonyms and one or more root words for each token in each of the plurality of text strings from a database;
for each of the plurality of text strings:
selecting, by the asset analysis module, a first text string from the plurality of text strings;
determining, by the asset analysis module, the number of matching tokens in the first text string and the unselected text strings based on the number of tokens, the number of root words of the tokens, and the number of synonyms of the tokens;
dividing, by the asset analysis module, the number of matching tokens by the number of tokens in the first text string to determine an asset mix value for the first text string;
summing, by the asset analysis module, each of the asset mix values for each text string to calculate an overall asset mix value;
determining, by the asset analysis module, a number of categories for each of the plurality of text strings using a category determination model;
summing, by the asset analysis module, the number of categories for each of the plurality of text strings to calculate an overall categorical coverage value;
retrieving, by the asset analysis module, an estimated traffic value for each of the plurality of target keywords;
counting, by the asset analysis module, the number of occurrences of the selected target keyword in each of the plurality of text strings to generate a target keyword count associated with the respective target keyword;
multiplying, by the asset analysis module, each of the target keyword counts with the estimated traffic value of the respective target keyword to calculate a weighted keyword value associated with the respective target keyword;
summing, by the asset analysis module, each of the weighted keyword values to calculate a keyword coverage value;
determining, by an asset scoring module, an overall content item score using a content scoring model, based on the number of text strings, the number of tokens in the text strings, the overall asset mix value, the overall categorical coverage value, and the overall keyword coverage value; and
providing, by the asset scoring module to the content provider computing device, the overall content item score associated with the plurality of text strings and the plurality of target keywords.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
identifying, by the asset reception module, if the plurality of text strings exist at a location in the database, wherein the location is determined based on an identifier received from the content provider computing device; and
storing, by the asset reception module, the plurality of text strings at the location in the database responsive to determining the plurality of text strings do not exist in the database based on the location;

### Claim 3 (depends on 2)

The method of claim 2, further comprising:
storing, by the asset scoring module, the overall content item score in the database;
retrieving, by the asset scoring module, a plurality of overall content item scores from the database, each of the overall content item scores associated with a plurality of text strings;
sorting, by the asset scoring module, the plurality of overall content item scores to create a sorted list of content item scores;
providing, by the asset scoring module to the content provider computing device, the sorted list of content item scores;

### Claim 4 (depends on 1)

The method of claim 1, wherein summing the number of categories for each of the plurality of text strings to calculate an overall categorical coverage value further comprises:
comparing, by the asset analysis module, the sum of the number of categories for each of the plurality of text strings to a first predefined threshold to create a first relationship;
comparing, by the asset analysis module, the sum of the number of categories for each of the plurality of text strings to a second predefined threshold to create a second relationship; and
calculating, by the asset analysis module, the categorical coverage value based on the first relationship and the second relationship.

### Claim 5 (depends on 1)

The method of claim 1, summing each of the weighted keyword values to calculate the keyword coverage value further comprises:
receiving, by the asset reception module from the content provider computing device, a target location;
retrieving, by the asset analysis module, a plurality of location keywords associated with the target location;
counting, by the asset analysis module, the number of occurrences of each of the plurality of location keywords in each of the plurality of text strings to generate a location keyword count associated with the respective location keyword; and
summing, by the asset analysis module, each of the location keyword counts to calculate a location relevance value.

### Claim 6 (depends on 5)

The method of claim 5, wherein determining the overall content item score is based on the location relevance value.

### Claim 7 (depends on 1)

The method of claim 1, further comprising:
receiving, by the asset reception module from the content provider computing device, a first address, the first address associated with a landing page;
accessing, by the asset analysis module, the landing page associated with the first address to retrieve a plurality of landing page tokens; and
counting, by the asset analysis module, the number of tokens in the plurality of text strings that match the tokens in the plurality of landing page tokens to determine a landing page relevance value.

### Claim 8 (depends on 7)

The method of claim 7, wherein determining the overall content item score is based on the landing page relevance value.

### Claim 9 (depends on 7)

The method of claim 7, wherein determining the landing page relevance value further comprises:
determining, by the asset analysis module, whether the landing page is a home page or a multi-product page; and
determining, by the asset analysis module, the landing page relevance value based on the determination that the landing page is a home page or a multi-product page.

### Claim 10 (depends on 1)

The method of claim 1, wherein:
determining, the asset analysis module, a number of categories for each of the plurality of text strings is based on a recurrent neural network model; and
determining, the asset analysis module, the overall content item score using a content scoring model is based on a linear regression model.

### Claim 11

A system of one or more processors, the processors configured to:
receive, via a network, a plurality of text strings, each of the plurality of text strings comprising a token;
receive, via a network, a plurality of target keywords;
retrieve one or more synonyms and one or more root words for each token in each of the plurality of text strings from a database;
for each of the plurality of text strings:
select a first text string from the plurality of text strings;
determine the number of matching tokens in the first text string and the unselected text strings based on the number of tokens, the number of root words of the tokens, and the number of synonyms of the tokens;
divide the number of matching tokens by the number of tokens in the first text string to determine an asset mix value for the first text string;
sum each of the asset mix values for each text string to calculate an overall asset mix value;
determine a number of categories for each of the plurality of text strings using a category determination model;
sum the number of categories for each of the plurality of text strings to calculate an overall categorical coverage value;
retrieve an estimated traffic value for each of the plurality of target keywords;
count the number of occurrences of the selected target keyword in each of the plurality of text strings to generate a target keyword count associated with the respective target keyword;
multiply each of the target keyword counts with the estimated traffic value of the respective target keyword to calculate a weighted keyword value associated with the respective target keyword;
sum each of the weighted keyword values to calculate a keyword coverage value;
determine an overall content item score using a content scoring model, based on the number of text strings, the number of tokens in the text strings, the overall asset mix value, the overall categorical coverage value, and the overall keyword coverage value; and
provide, via a network, the overall content item score associated with the plurality of text strings and the plurality of target keywords.

### Claim 12 (depends on 11)

The system of claim 11, further configured to:
identify if the plurality of text strings exist at a location in the database, wherein the location is determined based on an identifier received from the content provider computing device; and
store the plurality of text strings at the location in the database responsive to determining the plurality of text strings do not exist in the database at the location.

### Claim 13 (depends on 12)

The system of claim 12, further configured to:
store the overall content item score in at the location in the database.
retrieve a plurality of overall content item scores from the database, each of the plurality of overall content item scores associated with a plurality of text strings;
sort the plurality of overall content item scores to create a sorted list of content item scores;
provide, to the content provider computing device, the sorted list of content item scores;

### Claim 14 (depends on 11)

The system of claim 11, further configured to:
compare the sum of the number of categories for each of the plurality of text strings to a first predefined threshold to create a first relationship;
compare the sum of the number of categories for each of the plurality of text strings to a second predefined threshold to create a second relationship; and
calculate the categorical coverage value based on the first relationship and the second relationship.

### Claim 15 (depends on 11)

The system of claim 11, further configured to:
receive, via a network, a target location;
retrieve a plurality of location keywords associated with the target location from the database;
count the number of occurrences of each of the plurality of location keywords in each of the plurality of text strings to generate a location keyword count associated with the respective location keyword; and
sum each of the location keyword counts to calculate a location relevance value.

### Claim 16 (depends on 15)

The system of claim 15, further configured to:
determine the overall content item score based on the location relevance value.

### Claim 17 (depends on 11)

The system of claim 11, further configured to:
receive, via a network, a first address, the first address associated with a landing page;
access the landing page associated with the first address to retrieve a plurality of landing page tokens; and
count the number of tokens in the plurality of text strings that match the tokens in the plurality of landing page tokens to determine a landing page relevance value.

### Claim 18 (depends on 17)

The system of claim 17, further configured to:
determine the overall content item score based on the location relevance value.

### Claim 19 (depends on 17)

The system of claim 17, further configured to:
determine whether the landing page is a home page or a multi-product page; and
calculate the landing page relevance value based on the determination that the landing page is a home page or a multi-product page.

### Claim 20 (depends on 11)

The system of claim 11, further configured to:
determine a number of categories for each of the plurality of text strings is based on a recurrent neural network model; and
determine the overall content item score using a content scoring model is based on a linear regression model.
