Identifying Users in Social Networks with Limited Information

Abstract

We study the problem of Entity Resolution (ER) with limited information. ER is the problem of identifying and merging records that represent the same real-world entity. In this paper, we focus on the resolution of a single node g from one social graph (Google+ in our case) against a second social graph (Twitter in our case). We want to find the best match for g in Twitter, by dynamically probing the Twitter graph (using a public API), limited by the number of API calls that social systems allow. We propose a strategy that is designed for limited information and is robust to rule changing. We evaluate our strategy against random on a real dataset and show that our strategy is 46% faster than random and achieves 88.3% accuracy. We also propose a greedy strategy, which is a natural extension of bipartite matching subject to limited API calls, and it achieves 85% accuracy.

Available Versions of this Item

• Identifying Users in Social Networks with Limited Information. (deposited 21 Feb 2014 14:34) [Currently Displayed]
  • Identifying Users in Social Networks with Limited Information. (deposited 10 Oct 2014 12:58)
    • Identifying Users in Social Networks with Limited Information. (deposited 28 Oct 2014 00:00)
      • Identifying Users in Social Networks with Limited Information. (deposited 04 Dec 2014 17:51)
        • Identifying Users in Social Networks with Limited Information. (deposited 03 Apr 2015 00:56)