I propose a new data structure that reduces data sizes for data in which two item types have many-to-many relations. The proposed data structure newly introduces container variables related to many values of both items, and these container variables record many-to-many relations between them. The proposed data structure maintains data normalization and integrity and is independent of indexing methods conventionally used for relational databases, allowing simultaneous use of both.
When one item type has N items and the other item type has M items and all of N is related to all of M, the conventional RDB requests N×M rows, whereas the proposed data structure requests N+M rows. When N=100,000 and M=10,000,000, the conventional RDB requests 1,000,000,000,000 rows, whereas the proposed data structure requests only 10,100,000 rows.
In detail, please show the journal article linked by
https://www.iaiai.org/journals/index.php/IEE/article/view/589
, or, please show the US Patent No.11294961.