OptMatch

This is the detail page of the paper
OptMatch: Optimized Matchmaking via Modeling the High-Order Interactions on the Arena
accepted by KDD2020 ADS Oral

>> KDD2020 paper webpage
>> Introduction video (3min)
>> Presentation video (16min) [slides]

The Datasets used in OptMatch

Datasets Statistics

Dataset Matches Heroes Players
Dota2 (5v5) 50,000 113 10,815
LOL (5v5) 623,263 145 /
LOL Championship (5v5) 187,588 139 43,706
NBA 3,342 / 949
Fever Basketball 851,648 40 33,873

The Industrial Dataset(Fever Basketball) contains the current matches in an Online Game. (Currently under the process of check before data release)

Data Analysis and Effectiveness Comparison

KEEP UPDATING

As the datasets are too big, the github repository contains only example data for you to try out the OptMatch. To download the full datasets, please go to the following links.

  1. Dota2
    • Detail page
    • Full data link: [Google Cloud] or [Baidu Cloud]
  2. LOL
  3. LOL Championship
  4. NBA
    • Detail page
    • Full data link: [2018 season] and [2019 season]
  5. Fever Basketball
    • Detail page
    • Full data link: we will update the link once it’s available