International Business Machines Corporation

Lars Bremer

Martin Oberhofer

Yannick Saillet

Michael Baessler

Mike W Grasselt

Albert Maier

Hemanth Kumar Babu

Thuany Karoline Stuart

Mariya Chkalova

Thomas A P Hampp-Bahnmueller

Markus Lorch

Pavlo Petrenko

Jonathan Roesner

Karin Steckler

Sebastian B Schmid

Soma Shekar Naganna

Abhishek Seth

Robert Kern

Jason Allen Woods

James Albert O'Neill, Jr

Basem Elasioty

Claudio Andrea Fanconi

Holger Koenig

Alexandre Luz Xavier Da Costa

Utkarsh Bajpai

Benjamin Fabian Hogl

In an approach, a processor receives a request of a document. A processor identifies a set of datasets comprising a sensitive dataset, the set of datasets being interrelated in accordance with a relational model. A processor extracts attribute values of the document. A processor determines that a set of one or more attribute values of the extracted attribute values is in the set of datasets, the set of attribute values being values of a set of attributes. A processor determines that one or more entities of the sensitive dataset can be identified based on relations of the relational model between the set of attributes, where at least part of attribute values of the one or more entities comprises sensitive information. A processor, responsive to determining that the one or more entities can be identified, masks at least part of the set of one or more attribute values in the document.

Protecting sensitive data in documents

Embodiments of the present invention provide methods, computer program products, and systems. Embodiments of the present invention can condense a hierarchy in a data governance system, wherein the hierarchy comprises a root node and at least one child node comprising related sub-trees by determining, for a parent node in the hierarchy of governance system, governance terms and respective assignment relationships from a plurality of information assets, determining usage of the governance term in at least one of a plurality of governance rules, and marking a governance term of the plurality of governance terms for elimination based on the determined assignment relationships and the determined usage of the governance term in the plurality of governance rules. Embodiments of the present invention can then delete the governance term from the hierarchy if the governance term is marked for elimination.

Condensing hierarchies in a governance system based on usage

A database system can comprise records, each record including a set of attributes. The database system can further comprise database views, each database view representing a subset of the set of attributes. Data purpose objects indicating a subset of attributes of the set of attributes and a processing purpose can be stored. Each processing purpose can be associated with one or more entities that authorized access to the subset of attributes of the processing purpose. A request for data for a specific processing purpose and a selected view of the database views can be received. A data purpose object that indicates the specific processing purpose can be retrieved. The subset of attributes represented by the selected view can be compared with the subset of the attributes indicated in the retrieved data purpose object. Values of the subset of attributes of the selected view can be provided.

Conditional access to data

The present disclosure relates to a method comprising providing a set of one or more records, each record of the set of records having a set of one or more attributes. Values of the set of attributes of the set of records may be input to a trained data representation learning model for receiving, as an output of the trained data representation model, a set of feature vectors representing the set of records respectively. The set of feature vectors may be stored.

Method and system for processing data records

The exemplary embodiments disclose a method, a computer program product, and a computer system for protecting sensitive information. The exemplary embodiments may include using an inverted text index for evaluating one or more statistical measures of an index token of the inverted text index, using the one or more statistical measures for selecting a set of candidate tokens, extracting metadata from the inverted text index, associating the set of candidate tokens with respective token metadata, tokenizing at least one document resulting in one or more document tokens, comparing the one or more document tokens with the set of candidate tokens, selecting a set of document tokens to be masked, selecting at least part of the set of document tokens that comprises sensitive information according to the associated token metadata, masking the at least part of the set of document tokens, and providing one or more masked documents.

Masking sensitive information in a document

Duplicating identification in a graph having a second version. The graph includes nodes representing entities and edges representing relationships between the entities. The method includes: providing a previous first version of the graph. At least two target nodes of the graph may be identified. The first version and the second version of the graph may be compared for determining a comparison score indicative of a similarity between the two target nodes. The comparison score may be used for determining whether the two target nodes are duplicates with respect to each other.

Duplicate determination in a graph using different versions of the graph

A computer implemented method comprising processing the unstructured objects of each record of records of a database for identifying a set of one or more values of attributes in the unstructured objects of the each record. The sets of unstructured attribute values of two records of the database may be compared for determining a similarity level between the two sets. It may be determined whether the two records are representing a same entity based on the comparison result.

Record matching in a database system

The present disclosure relates to a method for enabling data integration. The method comprises collecting matching results of matching of records by a matching component over a time window. The number of false tasks of user defined tasks and system defined tasks in the collected matching results may be determined. The matching criterion used by the matching component may be adjusted to minimize the number of user defined tasks while the fraction of false tasks stays within a certain limit. The matching criterion may be replaced by the adjusted matching criterion for further usage of the matching component.

Method for configuring a matching component

Provided is a technique for matching different user representations of a person in a plurality of computer systems may be provided. The technique includes collecting information sets about user representations from a plurality of computer systems; normalizing the information sets to a unified format; grouping the information sets in the unified format into indexing buckets based on a user name using a non-phonetic algorithm; determining a similarity score for each pair of information sets in each of the indexing buckets; classifying each information set pair into a set of classes based on the similarity scores, wherein the set of classes comprise at least matches and non-matches; and using a data structure for merging information of information set pairs classified as matches.

Custodian disambiguation and data matching

Embodiments of the present invention provide for a method for duplicate determination in a graph. The graph comprises nodes representing entities and edges representing relationships between the entities. Embodiments of the present invention identify two target nodes in the graph. A neighborhood subgraph may be determined for each of the two nodes. The neighborhood subgraph includes the respective node. Embodiments of the present invention can then determine whether the two nodes are duplicates with respect to each other, based on a result of a comparison between the two subgraphs and based on a comparison of the two target nodes of the subgraphs.

Duplicate determination in a graph

Embodiments of the present invention determines duplicates in a graph. The graph comprises nodes representing entities and edges representing relationships between the entities. The method comprises: identifying at least two nodes in the graph. A neighborhood subgraph may be determined for each of the two nodes. The neighborhood subgraph includes the respective node. The method further comprises determining whether the two nodes are duplicates with respect to each other, based on a result of a comparison between the two subgraphs.

Method for duplicate determination in a graph

A computer implemented method is used for sorting data elements of a given set. The method includes performing an evaluation of a first type of usage of each data element. The method includes determining a set of data element candidates dependent on the evaluation of the first type of usage. The method includes performing an evaluation of a second type of usage of each data element of the set of data element candidates. The method includes sorting the data elements of the set of data element candidates dependent on the evaluation of the second type of usage of each data element of the set of data element candidates. The method includes providing the sorted data elements of the set of data element candidates, and in response, receiving a request for a data processing based on the provided sorted data elements of the set of data element candidates.

Sorting data elements of a given set of data elements

A method includes training a machine learning model using a current set of labeled data points. Each of the data points is multiple data records. A label of a data point indicates a classification of the data point. The training results in a trained machine learning model configured to classify a data point as representing a same entity or different entities. The method includes selecting a subset of unlabeled data points from a current set of unlabeled data points using classification results of the current set of unlabeled data points. The method includes providing the subset of unlabeled data points to a classifier and in response to providing receiving labels of the subset of unlabeled data points. The method may be repeated using the subset of labeled data points in addition to the current set of labeled data points as the current set of labeled data points.

Active learning for data matching

Matching records in an entity resolution system by defining entity attribute feature vectors, determining an entity attribute matching score according to a distance between two entity attribute feature vectors, assigning a statistical weight to an entity attribute matching score, adjusting the entity attribute matching score according to the statistical weight and an entity attribute frequency of occurrence, and determining an aggregate entity attribute matching score.

Growing community of inventors

Boeblingen, Germany

Lars Bremer