Browse Prior Art Database

System and Method for Recommending Social Media Hashtags Disclosure Number: IPCOM000222352D
Publication Date: 2012-Sep-26
Document File: 3 page(s) / 22K

Publishing Venue

The Prior Art Database


Disclosed is a system for automatic hashtag recommendations for social media posts.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 01 of 3

System and Method for Recommending Social Media Hashtags

On social media websites, when users submit micro-blog posts, they have to manually identify hashtags (i.e., #) through trending, friends, etc. to append to their posts. Hashtags are important because they allow posts to be "categorized" based on the content of the post or the linked-to document. Because of the manual effort involved in hashtagging, this is often error-prone and leads to users incorrectly tagging their own posts, or otherwise specifying a less-than-optimal hashtag for their post.

Prior art found under this subject outlines a tag recommendation system for social networks. The major drawbacks for existing inventions are that they do not consider the final length of micro-blogging posts. For example, one application has a maximum post length of 140 characters, so if a user enters a post that is 130 characters, existing systems may find a tag that is 12 characters in length, and would have to either recommend a tag that is too long to fit in the post, or drop the recommendation altogether. Also, existing prior art does not consider the auto-correction of tags that are entered by the user, taking that user's social connections and activities into account (along with the content of the post and linked-to Uniform Resource Locator (URL)).

Due to the popularity of micro-blogging and social media applications/sites, a method is needed to overcome the challenges mentioned above.

The invention is a hashtag recommendation system that:

1. Analyze the content of the post and the linked document's content

2. Analyze the social connections and activities of the user

3. Identify common hashtags from other posts that have similar content as #1, placing importance on frequency of tag use

4. Identify common hashtags from other social connections' posts (from #2), placing importance on the degree of separation and frequency of tag use

5. Identify currently-trending (or "hot") hashtags for similar content

6. Extending #5 by grouping currently-trending hashtags together by region/geographical location (e.g., current weather or natural disasters)

7. Weighting the tags obtained from #3, #4, #5 and #6

8. If the user has not selected any hashtags to include in the post, then this system recommends tags from #7, performing the following before-hand:

A. If the length of the post would be exceeded for any of the recommended tags, the system attempts to find aliases/synonyms for those hashtags that would keep the final post length under the maximum character limit

B. The similar hashtags may be identified by looking at content overlap and social connections for all of the instances where the "lengthy" hashtag is found.

9. If the user has already selected hashtags, then the system first makes sure that identical hashtags exist. If not, the user is prompted with a list of auto-correction options for hashtags. This auto-correction system considers #8, as well as "leans" more toward hashtags...