Browse Prior Art Database

Tag Recommendation for Social Media Textual Content Based on Crowdsourcing and Filters

IP.com Disclosure Number: IPCOM000240866D
Publication Date: 2015-Mar-06
Document File: 6 page(s) / 193K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a process that uses a crowdsourcing approach for tagging social media content. This tag recommendation process uses several techniques, including filtering and customization, to provide the tags that are most relevant to the content.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 28% of the total text.

Page 01 of 6

Tag Recommendation for Social Media Textual Content Based on Crowdsourcing and Filters

With the growing popularity of community driven social media sites such as Facebook* and Twitter*, users share massive amounts of textual content with friends and global users with the world. Social media sites provide tags to categorize content and deliver efficiently deliver it to users. Tags are descriptive keywords that can also be used to query content. For example, an article about California may be tagged with #sunny, #beaches, #Disneyland, #SF.

When publishing content to a social media site, a user might have difficulty determining which tags are most relevant to the content. For example, new Twitter users who are posting the first few tweets may not know which tags are most descriptive, relevant, and popular to optimally reach the appropriate audience. In addition, business users or professional accounts might want to customize and prioritize tags for business purposes.

The novel contribution is a process that uses a crowdsourcing approach for tagging social media content. This tag recommendation process uses several techniques to provide the tags that are most relevant to the content. The first technique is to prioritize the tags through filters. The second technique is to allow advertisers to promote associated tags specific to the business. The process relies on crowdsourcing to help reinforce the tag recommendation process.

When a social media user in not certain which tags to use, the novel process recommends the tags that are most relevant to the content. This includes ranking the list of recommended tags based on how other users have tagged similar content, tag popularity, geography, age group, etc. The filters can be outsourced to a third party. In addition, the system can connect words from different languages to enable annotating content in one language with tags from another language. Sponsored tags can be uses to monetizing tag recommendations. The system also enables both content authors and audiences to impact tag popularity

This process and system for tag recommendation is based on crowdsourcing. It starts by processing the content that is already tagged to generate a mapping between content keywords and tags used. Each keyword is associated with a list of tags and each association is assigned a weight factor in the range [0, 1] to indicate how relevant the tag is to the keyword. The higher the weight is, the more relevant the tag is. A Natural Language Processing (NLP) Application

Programming Interface (API) can be used to consume the content and produce a set of keywords. Once this initial step is done, the system makes available a repository containing keywords/tag mapping indicating how relevant each associated tag is to a certain keyword. In addition, each tag has a set of associated users. Each of these users has a user profile that can be data mined. In addition, the system stores the user's profile history along...