Brand abuse monitoring system with infringement deteciton engine and graphical user interface

Patent metadata

Field	Value
Publication number	US-2016253679-A1
Application number	US-201615052595-A
Country	US
Kind code	A1
Filing date	Feb 24, 2016
Priority date	Feb 24, 2015
Publication date	Sep 1, 2016
Grant date	—

Abstract

A computer-based system connected to a remote user device and a plurality of services. A data collection engine having an input adapted to receive a set of collection data from a set of services, the set of collection data comprising a set of brand data for comparison against a control set of data stored in a database and having a set of fields and via the digital communications network a query. An entity identification module identifies and extracts within the set of collection data a set of entity data, compares the set of entity data against a set of entity control data to resolve entity identification, associates an entity identifier with the identified entity, and determines a set of relationship data between entities; a matching module adapted to receive data derived from the set of collection data and compare the received data against a set of known brand data and generate a set of comparison data and an output indicating a presence of a match; a scoring module adapted to generate a set of score data based on the set of comparison data. A graphic user interface presents a graphic representation of the set of scored data on a display device.

First claim

We claim:

1 . A computer-based system connected via a communications network to a remote user device and a plurality of services, the system comprising:
a data collection engine comprising: an input adapted to receive a set of collection data from a set of services, the set of collection data comprising a set of brand data for comparison against a control set of data stored in a database and having a set of fields and via the digital communications network a query; an entity identification module adapted to identify and extract within the set of collection data a set of entity data, compare the set of entity data against a set of entity control data to resolve entity identification, associate an entity identifier with the identified entity, and determine a set of relationship data between entities; a matching module adapted to receive data derived from the set of collection data and compare the received data against a set of known brand data and generate a set of comparison data and an output indicating a presence of a match; a scoring module adapted to generate a set of score data based on the set of comparison data; and a graphic user interface adapted to present a graphic representation of the set of scored data on a display device associated with the remote user device.

2 . The system of claim 1 , further comprising a search module adapted to rank by a first relevance metric directed to a brand protection criteria.

3 . The system of claim 1 , further comprising a smart group-by means for allowing bulk action based on a brand protection criteria.

4 . The system of claim 1 further comprising a classification module adapted to classify entities by brand abuse category.

5 . The system of claim 1 wherein further comprising a classification module adapted to transform text data extracted from the set of document data into numeric vectors.

6 . The system of claim 1 further comprising means for performing a text to vector transformation using one of a topic model or Latent Dirichlet Allocation (LDA).

7 . The system of claim 1 further comprising a classification module adapted to use the Random Forest (RF) algorithm.

8 . The system of claim 1 further comprising a graph layer adapted to identify relationships between entities into a directed property graph, wherein entities (vertices) and relations (edges) of the graph are multi-valued.

9 . The system of claim 1 further comprising an attribute matching module.

10 . The system of claim 1 wherein the data collection engine includes uses a cluster operating on a set of connected computers to obtain brand protection information by one or more of crawling web pages over the Internet, harvesting search engines, crawling online market places, collecting social media feeds, detecting potential phish emails and landing pages, collecting mobile application data.

11 . The system of claim 1 wherein the data collection engine includes collecting detection data sets representing a unit data point or record comprising a set of fields and values and storing the detection data sets.

12 . The system of claim 11 wherein the detection data sets comprises one or more of the following fields: text such as product name or brand; numbers such as price or brand or product identifiers; images such as logos or trademarks or photographs; video media; structured data such as a HTML page or source code.

13 . The system of claim 11 wherein the data collection engine stores the detection datasets in a relational database or a NoSQL database distributed in a cluster of computers connected over a network.

14 . The system of claim 1 wherein the entity identification module establishes an entity resolution layer and includes natural language processing, machine learning and image matching techniques to identify fields and patterns in collected data and categorize them using an ontology of entities relevant to brand protection.

15 . The system of claim 1 wherein the entity resolution layer is configured based on a defined type or kind of entity and data source.

16 . The system of claim 1 wherein the entity resolution layer parses entities from unstructured text using one or more of text pattern matching techniques, efficient algorithms for fuzzy text pattern matching, Baeza-Yates-Gonnet algorithm for single strings, and fuzzy Aho-Corasick algorithm for multiple string matching.

17 . The system of claim 1 wherein the entity identification module includes using supervised or unsupervised document classification techniques after transforming collected text documents into numeric vectors using one or more of: multiple string fuzzy text pattern matching algorithms; fuzzy Aho-Corasick algorithm; topic models; Latent Dirichlet Allocation (LDA); and Hierarchical Dirichlet Processes (HDP).

18 . The system of claim 1 wherein the entity identification module determines a set of relationship data between entities based on collected detection data sets by categorizing detection into one from the set consisting of: ecommerce websites; marketplace listings; social media posts; and paid search advertisements, and categorizing the detection fields into one of the kinds of entities consisting of: “Key” entities which uniquely define a detection; “Node” entities which correspond to fields that have the potential to create relations between detections; and “Property” entities which correspond to fields that are not expected to create meaningful relations between detections, such as a price.

19 . The system of claim 1 further comprising a graph layer adapted to identify relationships between entities into a directed property graph, wherein entities are vertices and relations are edges of the graph and are multi-valued.

20 . The system of claim 19 wherein the graph layer comprises a vertex defined by a unique identifying number and a label, and wherein the set of vertex identity numbers of the vertices has an outgoing edge.

21 . The system of claim 20 wherein a graph is represented in a map structure where the vertex identity numbers are the keys and the list of other properties are the values.

22 . The system of claim 21 wherein the map structure is distributed in a cluster of computers using a distributed hashing technique.

23 . The system of claim 1 further comprising a cluster of computers connected over a network, the cluster adapted to load a brand dataset into a graph database and append a set of detection vertices to a graph vertex containing the value of a key node field appended to the graph.

24 . The system of claim 23 wherein property entity fields are appended as key/value pairs into a vertex created for the key entity of a current detection and wherein edges from the node vertices to the key vertices are appended to the graph.

25 . The system of claim 1 wherein collected data comprise images and further comprising an image matching and clustering module adapted to cluster images by determining a distance metric between a first image and a second image, the distance being determined by taking the mean squared pairwise difference of features extracted from each of the first image and the second image.

26 . The system of claim 25 further comprising a hierarchical clustering algorithm adapted to cluster images in the leaves of a prefix tree, wherein each node is represented by the average of a set of feature vectors of all images in chil

Assignees

• Thomson Reuters Global Resources

Inventors

• Venkatraman Krishna
• Gaudini Paul Garrett
• Hopping Joshua S
• Perreau Guimaraes Marcos
• Akella Prasad Narasimha

Classifications

• G06Q30/0185 Primary

  Product, service or business identity fraud · CPC title

• G06Q50/184

  Intellectual property management · CPC title