SCML 2021 League
One of the ANAC 2021 Competition Leagues
Days
Hours
Minutes
Seconds

NEWS

  • 2021.7.16 The fourth data release is available here.
  • 2021.7.15 We published the fifth SCML Newsletter today. If you did not receive it by email, you can find it here.
  • 2021.7.4 Deadline extension (see Important Dates).
  • 2021.6.30 The third data release is available here.
  • 2021.6.30 We published the fourth SCML Newsletter today. If you did not receive it by email, you can find it here. To receive future newsletters, please register here
  • 2021.6.19 As per the rules of the competitions, agents must have a non-trivial strategy to be considered for the official competition. As one of the mechanisms to enforce this rule, agents with no contracts in any simulations (most likely submitted to test the submission process) will not be ranked in the online competition and cannot become finalists in the official competition.
  • 2021.6.15 The second data release is now available.
  • 2021.6.14 We published the third SCML Newsletter today. If you did not receive it by email, you can find it here. To receive future newsletters, please register here
  • 2021.6.1 The first data release is now available.
  • 2021.6.1 We published the second SCML Newsletter today. If you did not receive it by email, you can find it here. To receive future newsletters, please register here
  • 2021.5.18 A new version of the SCML visualizer (0.2.4) was just published with bugfixes and a new tournament-only visualization of score distribution.
  • 2021.5.17 We published a blog post about SCML standard here.
  • 2021.5.10 We published the first SCML Newsletter today. If you did not receive it by email, you can find it here. To receive future newsletters, please register here
  • 2021.5.10 A new tutorial for the visualizer is now YouTube.
  • 2021.5.10 An updated version of scml and NegMAS are now on PyPi. Please upgrade your development environment.
  • 2021.5.2 The visualizer was improved substantially. You can now visualize contracts, negotiations, and offers. Moreover we added free data exploration to it. Stay tuned for the tutorial.
  • 2021.4.14 A visualizer for SCML was created. You can see a quick introduction about it here

This is the home of the Supply Chain Management League one of ANAC 2021 leagues @ IJCAI 2021.

The main motivation behind the SCM league is to increase the relevance of autonomous negotiation research by focusing on real-world scenarios that are characterized by situated negotiations, and complex dynamic utility functions.

The agent needs not only decide how to negotiate in a predefined single negotiation session, but when to join negotiations, and how to coordinate the behavior of its negotiators across multiple concurrent negotiations.

It is highly recommended to register your interest in the league by registering to this website by May 30th (No need to upload an agent by that time). It is mandatory to upload a preliminary version of the agent by July 10th.

Please note that the score on the live-competition will not impact in any way your final score in the official competition to be announced at IJCAI 2021.

The live competition is run using fewer configurations than the official competition to facilitate faster feedback when submitting new agents. This may lead to different results in the official competition.

Changes from SCML 2020

The main differences SCML 2020 and SCML 2021 can be found here.
  • Agents will have access to more information about the market to help them develop their strategies. These are specifically trading prices and exogenous contract summary (See Section 5 of the game description).
  • We introduce a simpler form of the game running in the same environment as a third track called SCML-OneShot. An overview of of this simpler form is found here. The full description of this simpler form is found here. The main differences between SCML-OneShot and the other two tracks can be found here.
  • New in May 10th, 2021 The truncated mean will be used instead of the median for comparing agents.
  • The evaluation criteria for the collusion track will be slightly modified to take into account a “the consolidated financial state” of all factories run by the same participant. More specifically, the total balance (and inventory) of all factories controlled by an agent type will be aggregated before calculating the profit giving a single score for the agent type in every simulation instead of three as was the case in 2020. This means that median-pumping by getting one factory to lose while pushing the profits of the other two up will not be effective in 2021.
  • Finalists for the “collusion” track MUST have a dedicated “collusion strategy” section in their reports with a non-trivial strategy for colluding among their factory managers.
  • We created discussion rooms for the three tracks here to help participants get together and contact organizers. You can also report bugs, feature requests and open new discussion topics on the same page.

Getting started and getting help

  1. The first thing you need to do is to read the game overview for the one shot track and/or the standard/collusion tracks.
  2. If you are not registered here, do so to received updates and be able to submit your agents later here.
  3. After deciding which track(s) to participate in, check there detailed description( OneShot Standard/Collusion).
  4. Next, Check the tutorials. You can find video tutorials and consult the documentation and we are happy to see you joining the discussions.
  5. Next, Download the appropriate skeleton (OneShot or standard/collusion) and test it. You will find a full description on how to do that in the README as well as the docstring on top of myagent.py in the skeleton .
  6. Happy hacking :-) If you have any questions, bug reports, feature requests, etc, you can open an issue on the SCML GitHub page

Sponsors

Links

Important Dates

Registration (Optional)May 30th
Preliminary SubmissionJuly 10th
Last Online TournamentJuly 28th
Final SubmissionJuly 31th
Report SubmissionAugust 10th
Finalist AnnouncementAugust 15th
Winner AnnouncementAugust

Statistics

Past Competitions

Leader Board (OneShot)

Based on the tournament run on 2021-07-26 17:37:47 UTC.

# Agent Score
1 UcOneshotAgent 9,923
2 UCOne 9,602
3 Agent74 9,575
4 Gentle 9,472
5 StagHunter 6,601
6 Zilberan 5,763
7 PDPSAgent 4,219
8 Agent112 -4,021
9 QlAgent -7,527
10 Agent97 -2,886,202
11 Godfather -2,971,919
12 BondAgent -8,104,563
13 TheSopranos78 -15,831,977
14 agentelfaroltest -2,479,768,070

Running next (14 Competitors): agentelfaroltest, StagHunter, Agent74, QlAgent, TheSopranos78, Zilberan, BondAgent, Godfather, Agent97, Gentle, UcOneshotAgent, Agent112, PDPSAgent, UCOne,

Leader Board (Standard)

Based on the tournament run on 2021-07-26 15:19:04 UTC.

# Agent Score
1 M4 467
2 KindestAgent 267
3 SteadyMgr (2020 winner) 247
5 BossAgent -126
6 Agent30 (2020 2nd place) -146
7 PhonyTales -149
8 PAIBIU -257
9 IyibiAgent -723
10 Esnaf007 -881
11 Agent68 -1,005
12 BlueWolf -1,095
13 Agent69 -1,164
14 AgentPerry -1,205
15 Perseverance -1,812
16 Agent73 -2,177
17 Agent137 -2,567
18 PA -2,807
19 Agent65 -2,850
20 YIYAgent -6,624
DursunAgent 0

Running next (20 Competitors): SteadyMgr, Agent30, M4, Agent65, KindestAgent, Agent68, Agent69, Agent73, PAIBIU, PA, Perseverance, Esnaf007, YIYAgent, Agent137, AgentPerry, IyibiAgent, DursunAgent, PhonyTales, BlueWolf, BossAgent,

Leader Board (Collusion)

Based on the tournament run on 2021-07-26 16:20:50 UTC.

# Team/Agent Score
1 M4 (2020 winner updated) 2,734
2 PhonyTales 1,482
3 BossAgent 496
4 Merchant (2020 2nd place) 224
6 KindestAgent -233
7 IyibiAgent -298
8 PAIBIU -370
9 Agent69 -429
10 BlueWolf -529
11 Agent68 -676
12 AgentPerry -841
13 Esnaf007 -859
14 Agent137 -1,431
15 PA -1,853
16 Agent65 -1,855
17 Perseverance -2,258
18 YIYAgent -2,289
19 Agent73 -2,375
DursunAgent 0

Running next (19 Competitors): M4, Merchant, Agent65, KindestAgent, Agent68, Agent69, Agent73, PAIBIU, PA, Perseverance, Esnaf007, YIYAgent, Agent137, AgentPerry, IyibiAgent, DursunAgent, PhonyTales, BlueWolf, BossAgent,

Sponsored by NEC-AIST AI Collaborative Research Laboratory and BIRD Initiative