Browse Prior Art Database

Influential group selection in the large-scale relational social analysis

IP.com Disclosure Number: IPCOM000236942D
Publication Date: 2014-May-22
Document File: 8 page(s) / 213K

Publishing Venue

The IP.com Prior Art Database

Abstract

Targeted customers selection is a fundamental problem in a marketing campaign. One of the objectives of selecting the targeted customers is to trigger a widespread adoption of innovations or merchandise. The cost of targeted marketing could be largely reduced if marketers have effectively chosen the target customers. However, traditionally proposed targeted customers selection techniques just rank the influence of all customers using some centrality metrics, and target the topmost customers. These traditional approaches run the risk of wasting resources on individual influencers, because some top influencers could have a large overlap on their influenced customers. In this idea, we propose a method and system to choose a set of influential groups instead of the individuals. This idea proposes a method for selecting a group of influential users in a social network. A time sensitive version of the PageRank is used to measure the importance of each user in the social network. A subset of most "important" or influential users is then chosen so as to maximize the coverage of users that can be influenced. The new aspect of the proposal lies in redundancy removal during the selection of a group of users rather than just selecting the top most influential users.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 16% of the total text.

Influential group selection in the large-scale relational social analysis

Targeted customers selection is a fundamental problem in a marketing campaign. One of the objectives of selecting the targeted customers is to trigger a widespread adoption of innovations or merchandise. The cost of targeted marketing could be largely reduced if marketers have effectively chosen the target customers. However, traditionally proposed targeted customers selection techniques just rank the influence of all customers using some centrality metrics, and target the topmost customers. These traditional approaches run the risk of wasting resources on individual influencers, because some top influencers could have a large overlap on their influenced customers. In this idea, we propose a method and system to choose a set of influential groups instead of the individuals.  This idea proposes a method for selecting a group of influential users in a social network. A time sensitive version of the PageRank is used to measure the importance of each user in the social network. A subset of most "important" or influential users is then chosen so as to maximize the coverage of users that can be influenced. The new aspect of the proposal lies in redundancy removal during the selection of a group of users rather than just selecting the top most influential users.

The idea proposes a novel approach to select influential groups. The heuristic influence maximization model utilizes a time-based PageRank centrality metric to measure the importance of each customer in spreading information. Further, it is able to select a group of seed influencers that could trigger the approximately maximum number of adoption of merchandise. In addition, it provides marketers with a portfolio-style strategy, enabling them to make use of the average effects of a group of influencers without relying on the seed influencers only. In the end, a one-month Twitter political campaign dataset is used to evaluate our model, reducing certain unrealistic assumptions about the cascade model.

Background:

Targeted customers selection has been studied in a number of domains, especially in viral marketing. One of the objectives of selecting the targeted customers is to trigger a widespread adoption of innovations or merchandise.   A lot of marketers use the strategy of paying customers who have sent emails or tweets recommending the companies’ merchandise to their friends, and the customers who buy the merchandise through the referred links will obtain discounts [1]. The incentive of this strategy is that social connection among customers is playing an increasing role of spreading innovations or merchandise nowadays. However, the influence of each customer is ignored; marketers just pay every customer who has successfully recommended their products.  In fact, customers have varying numbers of social connections and powers of persuading others to buy merchandise. The cost of targeted marketing will increase gr...