Browse Prior Art Database

System and Method to Classify Account Types of Social Media Authors

IP.com Disclosure Number: IPCOM000248589D
Publication Date: 2016-Dec-20
Document File: 4 page(s) / 39K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a system to automatically distinguish personal author accounts from organizational accounts in social media. The proposed system relies on a text analysis approach which extracts clues from (i) the authors’ profiles, including their biographies and names, and (ii) social media posts written by the authors.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 34% of the total text.

1

System and Method to Classify Account Types of Social Media Authors

Problem Description

Social Media Analysis solutions help customers understand what is being said about their products and services across social media sources such as Twitter, Facebook, Forums, Blogs or Review sites. Increasingly, it becomes important to not only identify what is being said, and what the sentiment towards the company is, but also who is saying it. For that matter, companies have come up with ways to detect influencers in social media, using a variety of algorithms. The goal of these algorithms is to provide customers with a set of authors (the influencers) that is worth reaching out to, and engage in 1:1 conversations to improve, for example, the perception of the customer’s brand.

The problem with these influencer lists is that they often contain authors that are, in fact, companies or news organizations. This limits the actionability of these influencer lists.

This proposal solves this problem by automatically classifying authors in social media into distinct categories. These include, but are not limited to:

• Organizations , for example https://twitter.com/Greenpeace

• Companies , for example https://twitter.com/Newco

• Media Channels , for example https://twitter.com/CNN

• Personal accounts, e.g., https://twitter.com/johndoe

Companies and Media Channels can be viewed as subtypes of the more general Organization .

With this presented idea, users of social media solutions can

1) Filter their influencer lists to only contain entries from personal accounts, which are much more amenable for “influencer marketing” and 1:1 conversations.

2) Break down the overall list of authors talking about their brand along these categories. This allows them to understand who is really driving the social media conversations around their brands and products: is it themselves (through official accounts), or is it really an “organic” conversation with lots of personal accounts involved.

Related Work

Related work is work that attempts to identify influencers in social media, or work that tries to elicit other attributes from social media authors, such as their gender or their marital status. Twitter also allows users to “verify” their accounts to become the “official” account of a certain person (often, a celebrity) or company.

There is also a body of work that attempts to classify spammers or bots in social media, essentially, accounts that spread harmful information. But to our knowledge, there has been no work to actually classify social media authors according to their

2

account type, and especially, trying to discern “real people” from company accounts.

Description of the Proposal

The core idea is to analyze the text that authors use in the description of their profile on the social media site. In Twitter, this is the so-called “author bio”, but similar information exists in review sites such as Amazon or TripAdvisor, or in video portals such as YouTube. Classification of...