Segmenting topical discussion themes from user-generated posts

Patent metadata

Field	Value
Publication number	US-10824660-B2
Application number	US-201514950550-A
Country	US
Kind code	B2
Filing date	Nov 24, 2015
Priority date	Nov 24, 2015
Publication date	Nov 3, 2020
Grant date	Nov 3, 2020

Abstract

Techniques are provided for detecting new topics and themes and assigning new posts to existing topic and/or theme clusters in online community discussions. A post posted to an online community is received and a post feature vector representative of the post is created. The post is compared to a plurality of centroid feature vectors, each centroid feature vector being representative of a respective post cluster and associated with a theme. Upon determining that similarity between the post feature vector and one of a plurality of centroid feature vectors satisfies a minimum similarity threshold, the post is assigned to the post cluster of which the centroid feature vector is representative. Upon determining that similarity between the post feature vector and any of the plurality of centroid feature vectors is below the minimum similarity threshold, a new theme cluster is created and the post is assigned to the new theme cluster.

First claim

What is claimed is:

1. One or more computer storage media storing computer-useable instructions that, when used by one or more computing devices, cause the one or more computing devices to perform operations comprising:
receiving a post;
creating a post feature vector representative of the post using a set of features that includes at least one text-based feature, and one or more of at least one time-based feature, wherein the at least one time-based feature is based upon time decay and at least one user-based feature, wherein the at least one user-based feature is based upon one of user interests and user participation;
determining whether similarity between the post feature vector and one of a plurality of centroid feature vectors satisfies a minimum similarity threshold, each of the plurality of centroid feature vectors being representative of a respective post cluster associated with a theme and including weighted entities derived from all posts comprising the respective post cluster;
upon determining that similarity between the post feature vector and one of the plurality of centroid feature vectors satisfies the minimum similarity threshold, assigning the post to the respective post cluster of which the one of the plurality of centroid feature vectors is representative; and
upon determining that similarity between the post feature vector and any one of the plurality of centroid feature vectors does not satisfy the minimum similarity threshold, creating a new post cluster associated with a new theme and assigning the post to the new post cluster.

2. A computerized theme segmenting engine comprising:
a user-generated post receiving component that receives a plurality of user-generated posts;
a vector creating component that creates a post feature vector representative of each received user-generated post using a set of features that includes at least one text-based feature, and one or more of at least one time-based feature, wherein the at least one time-based feature is based upon time decay, and at least one user-based feature, wherein the at least one user-based feature is based upon one of user interests and user participation;
a similarity assessing component that assesses the similarity between the post feature vector representative of each received user-generated post of the plurality of user-generated posts and a plurality of centroid feature vectors, each of the plurality of centroid feature vectors being representative of a respective post cluster associated with a theme and including weighted entities derived from all posts comprising the respective post cluster; and
a post assigning component that, upon determining that similarity between the post feature vector representative of one of the received user-generated posts of the plurality of user-generated posts and one of the plurality of centroid feature vectors satisfies a minimum similarity threshold, assigns the one of the received user-generated posts represented by the post feature vector representative of the one of the received user-generated posts to the respective post cluster represented by the one of the plurality of centroid feature vectors.

3. A computerized method for segmenting themes from user-generated posts, the computerized method comprising:
receiving, by a computing device, a user-generated post from an online community;
identifying, by the computing device, a plurality of entities within the user-generated post, each entity being representative of one feature of a set of features that includes at least one text-based feature, and one or more of at least one time-based feature, wherein the at least one time-based feature is based upon time decay, and at least one user-based feature, wherein the at least one user-based feature is based upon one of user interests and user participation;
assigning, by the computing device, a weight to each entity of the plurality of entities;
creating, by the computing device, a post feature vector based upon the weight assigned to each entity of the plurality of entities;
upon determining, by the computing device, that similarity between the post feature vector and any one of a plurality of centroid feature vectors is below a minimum similarity threshold, each of the plurality of centroid feature vectors being representative of a respective post cluster associated with a theme and including weighted entities derived from all posts comprising the respective post cluster, creating, by the computing device, a new post cluster associated with a new theme; and
assigning, by the computing device, the user-generated post to the new post cluster.

Assignees

• Adobe Inc

Inventors

• Jaidka Kokil
• Gupta Prakhar
• Rustagi Sajal
• Kaushik R

Classifications

• G06F16/35 Primary

  Clustering; Classification · CPC title