International Business Machines Corporation

Shlomo Steinhart

Benjamin Halberstadt

Ortal Nizri

Roger Kenneth Hecker

Erel Sharf

Boris Melamed

An example system includes a processor that can generate a first lineage graph based on a first set of monitored assets and processes used to produce a data asset. The processor can detect a data quality issue at the data asset. The processor can also generate a second lineage graph including a second set of monitored assets and processes that produced the data asset with the data quality issue. The processor can further compare the second lineage graph with the first lineage graph to detect a potential root cause of the data quality issue. The processor can also further modify an asset or process corresponding to the potential root cause of the data quality issue.

Detecting potential root causes of data quality issues using data lineage graphs

According to an aspect, data asset reconstruction includes receiving a data lineage for a data asset, where the data lineage identifies a data source connected to the data asset by an intermediate process. It is determined whether the intermediate process can be used to reconstruct a lost data asset. Based on determining that the intermediate process can be used to reconstruct the lost data asset, reconstructing the lost data asset with the intermediate process and the data source.

Data asset reconstruction

Transform governance rules for a data asset to apply to a set of related data assets. Establishing a governance rule over a first data asset based on a second governance rule applied to either an upstream or a downstream data asset.

Establishing governance rules over data assets

Creating data tables by identifying paths in a tree of data entities and relationships, the tree including data entities and relationships between the entities, where each of the paths originates from the same data entity, includes a terminal data entity that is at the to-many end of a one-to-many or many-to-many relationship with another data entity along the path, and includes no other of the data entities that lie along any path descending from the terminal data entity and that is at the to-many end of a one-to-many or many-to-many relationship with another of the data entities that lie along the path descending from the terminal data entity, and creating on a computer-readable medium a separate table for each of the paths, where each of the tables includes a column per each of the data entities of the path for which the table was created.

Growing community of inventors

Ramat Bet Shemesh, Israel

Shlomo Steinhart