Automated curation of digital information and user interface for same

Patent metadata

Field	Value
Publication number	US-11475485-B2
Application number	US-201816026117-A
Country	US
Kind code	B2
Filing date	Jul 3, 2018
Priority date	Jul 3, 2018
Publication date	Oct 18, 2022
Grant date	Oct 18, 2022

Abstract

A method of configuring a widget defining a user interface includes parsing content from a corpus comprising documents describing a plurality of products, selecting a plurality of pillars from the content, wherein the pillars are descriptors describing a product space of the products, determining an affinity of each of the products to each of the pillars, receiving, via the widget, a selection of a first pillar, and displaying a given product in the widget, the given product selected from the products given the selection of the first pillar.

First claim

What is claimed is:

1. A computer implemented method of configuring a widget running on a device and defining a graphical user interface of a cognitive advertisement, the method comprising:
accessing a corpus comprising documents describing a plurality of products;
parsing each of the documents within the corpus to identify the plurality of products and identify one or more modifiers for each of the plurality of products;
creating a semantic matrix comprising a plurality of cells populated with a frequency count of each of the modifiers for each of the plurality of products, wherein the semantic matrix includes a document vector for each product of the plurality of products, and each of the document vectors is constructed from the modifiers of a respective product of the plurality of products;
identifying, using the semantic matrix, a plurality of pillars from among the modifiers, wherein the plurality of pillars define a vector space of the plurality of products;
constructing a reverse-lookup dictionary of semantic distances from a plurality of positions within the vector space defined by the plurality of pillars to the document vectors of the plurality of products;
displaying the graphical user interface of the widget graphically presenting the vector space among the plurality of pillars, and a graphical user interface element configured for making a selection within the vector space corresponding to a relative selection among the plurality of pillars, wherein the semantic distances are translated into physical distances in the graphical presentation of the vector space;
receiving, via the graphical user interface element, a selection of a first position within the graphical presentation of the vector space associated with a particular combination of the plurality of pillars corresponding to the physical distances between the first position and each pillar of the plurality of pillars in the graphical user interface;
matching, using the reverse-lookup dictionary, a respective document vector of a first product of the plurality of products to the first position; and
displaying, based on the matching, the first product in the widget,
wherein identifying the plurality of pillars from among the modifiers comprises:
identifying a set of the modifiers in the semantic matrix appearing at least a given number of times in the corpus, wherein the set of the modifiers appearing at least the given number of times are candidate pillars;
scoring the candidate pillars based on the semantic distances to the products, wherein the semantic distances are cosine distances calculated between vectors representing the candidate pillars and the document vectors of the plurality of products; and
selecting the plurality of pillars from among the candidate pillars using the scoring, wherein the plurality of pillars differentiate, based on the scoring, among the plurality of products in the vector space.

2. The computer implemented method of claim 1 , wherein the parsing comprises at least one of a natural language parsing and an imaging parsing.

3. The computer implemented method of claim 1 , further comprising receiving at least one constraint on the graphical user interface of the widget.

4. The computer implemented method of claim 1 ,
further comprising:
determining, from a sentiment dictionary, a sentiment attribute for at least one of the modifiers; and
adding the sentiment attribute to the at least one of the modifiers.

5. A non-transitory computer readable medium comprising computer executable instructions which when executed by a computer cause the computer to perform a method of configuring a widget defining a graphical user interface of a cognitive advertisement, the method comprising:
accessing a corpus comprising documents describing a plurality of products;
parsing each of the documents within the corpus to identify the plurality of products and identify one or more modifiers for each of the plurality of products;
creating a semantic matrix comprising a plurality of cells populated with a frequency count of each of the modifiers for each of the plurality of products, wherein the semantic matrix includes a document vector for each product of the plurality of products, and each of the document vectors is constructed from the modifiers of a respective product of the plurality of products;
identifying, using the semantic matrix, a plurality of pillars from among the modifiers, wherein the plurality of pillars define a vector space of the plurality of products;
constructing a reverse-lookup dictionary of semantic distances from a plurality of positions within the vector space defined by the plurality of pillars to the document vectors of the plurality of products;
displaying the graphical user interface of the widget graphically presenting the vector space among the plurality of pillars, and a graphical user interface element configured for making a selection within the vector space corresponding to a relative selection among the plurality of pillars, wherein the semantic distances are translated into physical distances in the graphical presentation of the vector space;
receiving, via the graphical user interface element, a selection of a first position within the graphical presentation of the vector space associated with a particular combination of the plurality of pillars corresponding to the physical distances between the first position and each pillar of the plurality of pillars in the graphical user interface;
matching, using the reverse-lookup dictionary, a respective document vector of a first product of the plurality of products to the first position; and
displaying, based on the matching, the first product in the widget,
wherein identifying the plurality of pillars from among the modifiers comprises:
identifying a set of the modifiers in the semantic matrix appearing at least a given number of times in the corpus, wherein the set of the modifiers appearing at least the given number of times are candidate pillars;
scoring the candidate pillars based on the semantic distances to the products, wherein the semantic distances are cosine distances calculated between vectors representing the candidate pillars and the document vectors of the plurality of products;
selecting the plurality of pillars from among the candidate pillars using the scoring, wherein the plurality of pillars differentiate, based on the scoring, among the plurality of products in the vector space.

6. The computer readable medium of claim 5 , wherein the parsing comprises at least one of a natural language parsing and an imaging parsing.

7. The computer readable medium of claim 5 , further comprising receiving at least one constraint on the graphical user interface of the widget.

8. The computer readable medium of claim 5 ,
further comprising:
determining, from a sentiment dictionary, a sentiment attribute for at least one of the modifiers; and
adding the sentiment attribute to the at least one of the modifiers.

Assignees

• IBM

Inventors

• Byron Donna K
• Seabolt Edward E
• Swift Mary D
• Hull Brian

Classifications

• G06V20/00

  Scenes; Scene-specific elements (control of digital cameras H04N23/60) · CPC title

• G06F18/28

  Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries · CPC title

• G06F18/2413

  based on distances to training or reference patterns · CPC title

• G06Q30/0277

  Online advertisement · CPC title

• G06Q30/0643

  graphically representing goods, e.g. 3D product representation · CPC title