Organizational logo enrichment

Patent metadata

Field	Value
Publication number	US-10002292-B2
Application number	US-201514929116-A
Country	US
Kind code	B2
Filing date	Oct 30, 2015
Priority date	Sep 30, 2015
Publication date	Jun 19, 2018
Grant date	Jun 19, 2018

Abstract

In an example embodiment, a web page is obtained using a web page address stored in a first record and is parsed to extract one or more images from the web page along with a second plurality of features for each of the one or more images from the web page. Information about each image of the web page and the extracted second plurality of features for the web page are input into a supervised machine learning classifier to calculate a logo confidence score for each image of the web page, the logo confidence score indicating the probability that the image is an organization logo. In response to a particular image in the web page having a logo confidence score transgressing a first threshold, the particular image is injected into an organization logo field of the first record.

First claim

What is claimed is:

1. A computer-implemented method for enrichment of a logo field of a first record in a computer system, the method comprising:
retrieving a plurality of sample web pages;
for each sample web page in the plurality of sample web pages, parsing the sample web page to extract one or more images from the sample web page along with a first plurality of features for each of the one or more images from the sample web page, the plurality of features including proximity of each of the one or more images from the sample web page to a top of the sample web page;
labeling at least one of the images in one of the plurality of sample web pages as being an organization logo;
feeding information about the labeled at least one of the images, the extracted first plurality of feature for each sample web page into a supervised machine learning classifier to train the supervised machine learning classifier to recognize when an image in a web page is an organization logo, wherein the supervised machine learning classifier utilizes a random forest algorithm and a k-nearest neighbor classifier;
obtaining a web page using a web page address stored in the first record;
parsing the web page to extract one or more images from the web page along with a second plurality of features for each of the one or more images from the web page, the first plurality of features including proximity of each of the one or more images from the web page to a top of the web page, without rendering the web page;
inputting information about each image of the web page and the extracted second plurality of features for the web page into the supervised machine learning classifier to calculate a logo confidence score for each image of the web page, the logo confidence score indicating a probability that the image is an organization logo; and
in response to a particular image in the web page having a logo confidence score transgressing a first threshold, injecting the particular image into an organization logo field of the first record.

2. The method of claim 1 , wherein the first and second plurality of features both include image dimension.

3. The method of claim 1 , wherein the first and second plurality of features include whether an image links to a home page of an organization web site.

4. The method of claim 1 , wherein the second plurality of features include similarity of an organization name extracted from the first record to a file name of an image.

5. The method of claim 1 , wherein the second plurality of features include similarity of an organization name extracted from the first record to tag attributes surrounding a tag in the web page.

6. The method of claim 1 , further comprising feeding a vector of common terms either positively or negatively correlated with a logo into the supervised machine learning classifier and using the presence of any of these terms in a tag in the web page to influence whether a nearby image corresponds to an organization logo.

7. A system comprising:
a computer-readable medium having instructions stored there on, which, when executed by a processor, cause the system to perform operations comprising:
retrieving a plurality of sample web pages;
for each sample web page in the plurality of sample web pages, parsing the sample web page to extract one or more images from the sample web page along with a first plurality of features for each of the one or more images from the sample web page, the plurality of features including proximity of each of the one or more images from the sample web page to a top of the sample web page;
labeling at least one of the images in one of the plurality of sample web pages as being an organization logo;
feeding information about the labeled at least one of the images, the extracted first plurality of feature for each sample web page into a supervised machine learning classifier to train the supervised machine learning classifier to recognize when an image in a web page is an organization logo, wherein the supervised machine learning classifier utilizes a random forest algorithm and a k-nearest neighbor classifier;
obtaining a web page using a web page address stored in the first record;
parsing the web page to extract one or more images from the web page along with a second plurality of features for each of the one or more images from the web page, the first plurality of features including proximity of each of the one or more images from the web page to a top of the web page, without rendering the web page;
inputting information about each image of the web page and the extracted second plurality of features for the web page into the supervised machine learning classifier to calculate a logo confidence score for each image of the web page, the logo confidence score indicating a probability that the image is an organization logo; and
in response to a particular image in the web page having a logo confidence score transgressing a first threshold, injecting the particular image into an organization logo field of the first record.

8. The system of claim 7 , wherein the first and second plurality of features both include image dimension.

9. The system of claim 7 , wherein the first and second plurality of features include whether an image links to a home page of an organization web site.

10. The system of claim 7 , wherein the second plurality of features include similarity of an organization name extracted from the first record to a file name of an image.

11. The system of claim 7 , wherein the second plurality of features include similarity of an organization name extracted from the first record to tag attributes surrounding a tag in the web page.

12. The system of claim 7 , wherein the operations further comprise feeding a vector of common terms either positively or negatively correlated with a logo into the supervised machine learning classifier and using the presence of any of these terms in a tag in the web page to influence whether a nearby image corresponds to an organization logo.

13. A non-transitory machine-readable storage medium comprising instructions, which when implemented by one or more machines, cause the one or more machines to perform operations comprising:
retrieving a plurality of sample web pages;
for each sample web page in the plurality of sample web pages, parsing the sample web page to extract one or more images from the sample web page along with a first plurality of features for each of the one or more images from the sample web page, the plurality of features including proximity of each of the one or more images from the sample web page to a top of the sample web page;
labeling at least one of the images in one of the plurality of sample web pages as being an organization logo;
feeding information about the labeled at least one of the images, the extracted first plurality of feature for each sample web page into a supervised machine learning classifier to train the supervised machine learning classifier to recognize when an image in a web page is an organization logo, wherein the supervised machine learning classifier utilizes a random forest algorithm and a k-nearest neighbor classifier;
obtaining a web page using a web page address stored in the first record;
parsing the web page to extract one or more images from the web page along with a second plurality of features for each of the one or more images from the web page, the first plurality of features including proximity of each of the one or more images from the web page to a top of the web page, without rendering the web page;
inputting information about each image of the web page and the extracted second plurality of features

Assignees

• Microsoft Technology Licensing Llc

Inventors

• Guo Songtao
• Degiere Christopher Matthew
• Huang Jingjing
• Kumar Aarti
• Lai Alex Ching
• Li Xian

Classifications

• G06F16/215 Primary

  Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors · CPC title

• G06V30/19167

  Active pattern learning · CPC title

• G06V30/19173

  Classification techniques · CPC title

• G06V30/19147

  Obtaining sets of training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06V10/761

  Proximity, similarity or dissimilarity measures · CPC title