These are source codes of the paper under review.

• CODA are algorithms for counting directed hypergraphlets which have two versions.
  • CODA-E is a brute-force accurate algorithm.
  • CODA-A is an approximate algorithm based on weighted sampling.
• As baselines, we also propose approximate algorithms D-MoCHy and A2A sampling.
  • D-MoCHy is a direct extension of MoCHy-A+, the most advanced algorithm presented in "Hypergraph Motifs: Concepts, Algorithms, and Discoveries (Geon Lee, VLDB'20)" in that both methods involve uniformly sampling a pair of incident hyperedges using a projected graph.
  • A2A sampling is an approximate algorithm that samples a single hyperarc $e$ from $E$ uniformly at random first, and then another incident hyperarc $e^′ ∈ N_e$ uniformly at random.
• CODA-A has the advantage of time, memory, and accuracy over baselines and CODA-E.

• Statistics and references for each dataset are as follows.
• |V| and |E| are the number of nodes and hyperarcs, respectively.
• We provide processed data with .txt files.

• Command "bash src/run.sh".
  • datas is the list of data that are counted.
  • version is the list of counting algorithms.
  • fs is the list of fractions q where the number of samples becomes n=q|E|.
  • try is the number of iterations in counting.

<example>
datas="citation_data_science citation_software email_enron email_eu metabolic_iAF1260b metabolic_iJO1366 qna_math qna_server bitcoin_2014 bitcoin_2015 bitcoin_2016"
versions="CODAE CODAA A2A DMOCHY" 
fs="10"
try=3

1. (txt) the count of each directed hypergraphlet whose line is composed of ({index of a directed hypergraphlet},{counts})
2. (csv) time taken for reading and counting

• We provide a code implemented with C++ for obtaining triad census.
• command "bash triad/run.sh".