Results of the First Annual Human-Agent League of the Automated Negotiating Agents Competition

Abstract

We present the results of the first annual Human-Agent League of ANAC. By introducing a new human-agent negotiating platform to the research community at large, we facilitated new advancements in human-aware agents. This has succeeded in pushing the envelope in agent design, and creating a corpus of useful human-agent interaction data. Our results indicate a variety of agents were submitted, and that their varying strategies had distinct outcomes on many measures of the negotiation. These agents approach the problems endemic to human negotiation, including user modeling, bidding strategy, rapport techniques, and strategic bargaining. Some agents employed advanced tactics in information gathering or emotional displays and gained more points than their opponents, while others were considered more "likeable" by their partners.

The Challenge of Negotiation in the Game of Diplomacy

“…In each game the players were randomly assigned to the 7 Powers. 13 Every round of each game had a deadline of 30 seconds. In order to prevent the games from continuing forever a draw was declared automatically in any game that advanced to the Winter 1920 phase.…”

“…The setup of this competition has been updated each year to reflect the advancements made in the field of research. While ANAC started with small contract spaces and linear utility functions [4], it has featured increasingly complex scenarios, involving very large agreement spaces [8], multilateral negotiations [1], human-agent interactions [13], and non-linear utility functions [8,2].…”

“…https://pypi.python.org/pypi/Parlance/1.4.113 It would have been better to assign each agent to each Power an equal number of times, because some Powers are stronger than others. Unfortunately, however, the Parlance game server does not provide this option.…”

The Likeability-Success Tradeoff: Results of the 2nd Annual Human-Agent Automated Negotiating Agents Competition

“…In analysis of the competition data, we discover that the previously reported tradeoff between agent success and agent likeability [10] is more complex than at first assumed. Critical to negotiation, and to this important relationship, is the idea of "integrative potential".…”

“…The IAGO Negotiation platform was proposed and designed by Mell et al and was selected to be used for the Human-Agent League of ANAC [13] in 2017. Results from this first human-agent competition have been previously published [10]. IAGO provides a front-facing GUI for the human-participants (see Fig.…”