Patents by Inventor Jens P. Seifert
Jens P. Seifert has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Patent number: 11429878Abstract: A method, computer system, and computer program product for providing recommendations about processing datasets. A set of machine learning models are provided for use in respectively determining data processing action performable on a dataset based on a respective set of features of the dataset. A current dataset is received. A set of features of the current dataset are determined. One or more data processing actions are generated to be executed on the current dataset, which are determined by at least two machine learning models of the provided set, based on the determined set of features of the current dataset. One or more of the data processing actions are performed on the current dataset.Type: GrantFiled: September 22, 2017Date of Patent: August 30, 2022Assignee: International Business Machines CorporationInventors: Yannick Saillet, Martin A. Oberhofer, Jens P. Seifert
-
Patent number: 11023483Abstract: Embodiments of the present invention disclose generating a data profiling jobs for source data in a data processing system, the source data being described by at least one source functional data model. A target functional data model is provided, for describing target data that can be generated from the source data. One or more source functional data models are identified that correspond to the target functional data model. At least one functional source-to-target model mapping is associated to at least one source-target pair based on the target functional data model and identified source functional data models. A physical source-to-target model mapping for at least one source-target pair based on the logical source-to-target model mapping is calculated. For all physical source attributes, the needed data profiling jobs are generated based on the target attribute for analyzing the physical source attributes.Type: GrantFiled: August 4, 2016Date of Patent: June 1, 2021Assignee: International Business Machines CorporationInventors: Sebastian Nelke, Martin Oberhofer, Yannick Saillet, Jens P. Seifert
-
Patent number: 11023484Abstract: Embodiments of the present invention disclose generating a data profiling jobs for source data in a data processing system, the source data being described by at least one source functional data model. A target functional data model is provided, for describing target data that can be generated from the source data. One or more source functional data models are identified that correspond to the target functional data model. At least one functional source-to-target model mapping is associated to at least one source-target pair based on the target functional data model and identified source functional data models. A physical source-to-target model mapping for at least one source-target pair based on the logical source-to-target model mapping is calculated. For all physical source attributes, the needed data profiling jobs are generated based on the target attribute for analyzing the physical source attributes.Type: GrantFiled: December 6, 2017Date of Patent: June 1, 2021Assignee: International Business Machines CorporationInventors: Sebastian Nelke, Martin Oberhofer, Yannick Saillet, Jens P. Seifert
-
Patent number: 10970173Abstract: A logging process in a data storage system having a set of storage tiers, each storage tier of the set of storage tiers having different performance characteristics, wherein the set of storage tiers is divided into a plurality of subsets of storage tiers using the performance characteristics, may include initiating the logging process for creating a separate log file for each of the plurality of subsets of storage tiers for maintaining a history of data changes in the subset of storage tiers, thereby creating a plurality of log files. In response to a change in data stored in at least one storage tier of a subset of storage tiers of the plurality of subsets of storage tiers, one or more log records including information about the change may be generated and written into respective log files.Type: GrantFiled: November 30, 2018Date of Patent: April 6, 2021Assignee: INTERNATIONAL BUSINESS MACHINES CORPORATIONInventors: Martin Oberhofer, Jens P. Seifert, Kostas Rakopoulos, Stephen Rees
-
Patent number: 10740098Abstract: A method, computer program product, and computer system for providing a comparison result vector of a predefined number of elements w resulting from comparison of multiple vectors of compressed data within a processor comprising registers of same size m is provided. Vector elements of the comparison result vector are stored in a register of the registers. Zero bits are padded between vector elements of each of the comparison result vectors. A compare bit result vector indicative of the vector elements is generated for accessing the results of the comparison in the comparison result vector.Type: GrantFiled: February 6, 2018Date of Patent: August 11, 2020Assignee: International Business Machines CorporationInventors: Cedric Lichtenau, Silvia M. Mueller, Jens P. Seifert, Jörg-Stephan Vogt, Markus Lachenmayr, L'Emir Salim Chehab, Pavankrishna Ellore Ramesh, Sourabh Chougule
-
Publication number: 20200225941Abstract: The present disclosure relates to a method for creating run-time executables for data analysis functions. The method comprises in response to receiving a data analysis request from a user, selecting from a repository a repository of data analysis functions a set of data analysis functions for execution in a hosting environment or on premises of the user. Usage conditions of the set of data analysis functions by the user may be determined. An additional code for applying the determined usage conditions may be created. The selected data analysis functions and the additional code may be compiled, resulting in an executable code. The executable code may be certified. The certified executable code may be deployed or provided for download to a run-time environment for certified executable codes.Type: ApplicationFiled: January 15, 2019Publication date: July 16, 2020Inventors: Martin Oberhofer, Mike W. Grasselt, Yannick Saillet, Jens P. Seifert
-
Publication number: 20200225942Abstract: The present disclosure relates to a method for creating run-time executables for data analysis functions. The method comprises in response to receiving a data analysis request from a user, selecting from a repository a repository of data analysis functions a set of data analysis functions for execution in a hosting environment or on premises of the user. Usage conditions of the set of data analysis functions by the user may be determined. An additional code for applying the determined usage conditions may be created. The selected data analysis functions and the additional code may be compiled, resulting in an executable code. The executable code may be certified. The certified executable code may be deployed or provided for download to a run-time environment for certified executable codes.Type: ApplicationFiled: July 2, 2019Publication date: July 16, 2020Inventors: Martin Oberhofer, Mike W. Grasselt, Yannick Saillet, Jens P. Seifert
-
Patent number: 10621492Abstract: The present disclosure relates to a method for centrally processing data records using a record linkage algorithm. The method comprises providing a centralized master repository for storing data records in a predefined data structure having a set of attributes. At least one clustering metric is provided. Clusters of records may be determined using a clustering function that is based on the at least one clustering metric. For each particular cluster, a set of configuration data for the record linkage algorithm may be defined based on a value of the clustering metric within that particular cluster. The individual data records may be assigned to one or more clusters of the clusters using the clustering metric values and the record linkage algorithm may be applied to a set of two or more individual data records assigned to at least one common cluster using the set of configuration data for the common cluster.Type: GrantFiled: October 21, 2016Date of Patent: April 14, 2020Assignee: International Business Machines CorporationInventors: Martin Oberhofer, Yannick Saillet, Scott Schumacher, Jens P. Seifert
-
Patent number: 10621493Abstract: The present disclosure relates to a method for centrally processing data records using a record linkage algorithm. The method comprises providing a centralized master repository for storing data records in a predefined data structure having a set of attributes. At least one clustering metric is provided. Clusters of records may be determined using a clustering function that is based on the at least one clustering metric. For each particular cluster, a set of configuration data for the record linkage algorithm may be defined based on a value of the clustering metric within that particular cluster. The individual data records may be assigned to one or more clusters of the clusters using the clustering metric values and the record linkage algorithm may be applied to a set of two or more individual data records assigned to at least one common cluster using the set of configuration data for the common cluster.Type: GrantFiled: January 2, 2018Date of Patent: April 14, 2020Assignee: International Business Machines CorporationInventors: Martin Oberhofer, Yannick Saillet, Scott Schumacher, Jens P. Seifert
-
Patent number: 10614070Abstract: A method, computer program product, and computer system for optimizing query processing is provided. An asynchronously updated index is provided for a main dataset. A time-sequences log of data modifications to the main dataset is provided. A query of the main dataset is received. The main dataset is joined with the time-sequenced log data resulting in a first intermediate result. The query is processed by keeping one or more entries satisfying the query by emulating a function of the asynchronously updated index resulting in a second intermediate result. Updated, deleted dataset entries are deleted from the asynchronously updated index. The query is processed resulting in a third intermediate result. A union of the second intermediate result and third intermediate result is built defining a final result.Type: GrantFiled: October 27, 2015Date of Patent: April 7, 2020Assignee: International Business Machines CorporationInventors: Marion E. Behnen, Joern Klauke, Jens P. Seifert, Calisto P. Zuzarte
-
Patent number: 10606839Abstract: A method, computer program product, and computer system for optimizing query processing is provided. An asynchronously updated index is provided for a main dataset. A time-sequences log of data modifications to the main dataset is provided. A query of the main dataset is received. The main dataset is joined with the time-sequenced log data resulting in a first intermediate result. The query is processed by keeping one or more entries satisfying the query by emulating a function of the asynchronously updated index resulting in a second intermediate result. Updated, deleted dataset entries are deleted from the asynchronously updated index. The query is processed resulting in a third intermediate result. A union of the second intermediate result and third intermediate result is built defining a final result.Type: GrantFiled: May 23, 2017Date of Patent: March 31, 2020Assignee: International Business Machines CorporationInventors: Marion E. Behnen, Joern Klauke, Jens P. Seifert, Calisto P. Zuzarte
-
Patent number: 10579375Abstract: The present disclosure relates performing of comparisons between a first and a second vector. The memory location has a size or length of m bits. A compare block to compare two single bits is used. The compare block comprises: two input bits associated to one of the bits from the first and the second vector respectively; a greater than input bit and a lower than input bit; a cascade enable input bit to control if the greater than input bit and the lower than input bit are considered; a greater than result bit, a lower than result bit, and an equal result bit. A daisy chaining of m of the one-bit compare blocks is performed such that the result bits of one compare block represents the compare result of the previous compare blocks in the chain.Type: GrantFiled: February 6, 2018Date of Patent: March 3, 2020Assignee: International Business Machines CorporationInventors: Cedric Lichtenau, Silvia M. Mueller, Jens P. Seifert, Jörg-Stephan Vogt, Markus Lachenmayr, L'Emir Salim Chehab, Pavankrishna Ellore Ramesh, Sourabh Chougule
-
Publication number: 20190243649Abstract: The present disclosure relates a method, computer program product, and computer system to provide a comparison result vector of a predefined number of elements w resulting from comparison of multiple vectors of compressed data within a processor comprising registers of same size m. Vector elements of the comparison result vector are stored in a register of the registers. Zero bits are padded between vector elements of each of the comparison result vectors. A compare bit result vector indicative of the vector elements is generated for accessing the results of the comparison in the comparison result vector.Type: ApplicationFiled: February 6, 2018Publication date: August 8, 2019Inventors: Cedric Lichtenau, Silvia M. Mueller, Jens P. Seifert, Jörg-Stephan Vogt, Markus Lachenmayr, L'Emir Salim Chehab, Pavankrishna Ellore Ramesh, Sourabh Chougule
-
Publication number: 20190243650Abstract: The present disclosure relates performing of comparisons between a first and a second vector. The memory location has a size or length of m bits. A compare block to compare two single bits is used. The compare block comprises: two input bits associated to one of the bits from the first and the second vector respectively; a greater than input bit and a lower than input bit; a cascade enable input bit to control if the greater than input bit and the lower than input bit are considered; a greater than result bit, a lower than result bit, and an equal result bit. A daisy chaining of m of the one-bit compare blocks is performed such that the result bits of one compare block represents the compare result of the previous compare blocks in the chain.Type: ApplicationFiled: February 6, 2018Publication date: August 8, 2019Inventors: Cedric Lichtenau, Silvia M. Mueller, Jens P. Seifert, Jörg-Stephan Vogt, Markus Lachenmayr, L'Emir Salim Chehab, Pavankrishna Ellore Ramesh, Sourabh Chougule
-
Publication number: 20190102259Abstract: A logging process in a data storage system having a set of storage tiers, each storage tier of the set of storage tiers having different performance characteristics, wherein the set of storage tiers is divided into a plurality of subsets of storage tiers using the performance characteristics, may include initiating the logging process for creating a separate log file for each of the plurality of subsets of storage tiers for maintaining a history of data changes in the subset of storage tiers, thereby creating a plurality of log files. In response to a change in data stored in at least one storage tier of a subset of storage tiers of the plurality of subsets of storage tiers, one or more log records including information about the change may be generated and written into respective log files.Type: ApplicationFiled: November 30, 2018Publication date: April 4, 2019Inventors: Martin Oberhofer, Jens P. Seifert, Kostas Rakopoulos, Stephen Rees
-
Publication number: 20190095801Abstract: A method, computer system, and computer program product for providing recommendations about processing datasets. A set of machine learning models are provided for use in respectively determining data processing action performable on a dataset based on a respective set of features of the dataset. A current dataset is received. A set of features of the current dataset are determined. One or more data processing actions are generated to be executed on the current dataset, which are determined by at least two machine learning models of the provided set, based on the determined set of features of the current dataset. One or more of the data processing actions are performed on the current dataset.Type: ApplicationFiled: September 22, 2017Publication date: March 28, 2019Inventors: Yannick Saillet, Martin A. Oberhofer, Jens P. Seifert
-
Patent number: 10176049Abstract: A logging process in a data storage system having a set of storage tiers, each storage tier of the set of storage tiers having different performance characteristics, wherein the set of storage tiers is divided into a plurality of subsets of storage tiers using the performance characteristics, may include initiating the logging process for creating a separate log file for each of the plurality of subsets of storage tiers for maintaining a history of data changes in the subset of storage tiers, thereby creating a plurality of log files. In response to a change in data stored in at least one storage tier of a subset of storage tiers of the plurality of subsets of storage tiers, one or more log records including information about the change may be generated and written into respective log files.Type: GrantFiled: July 7, 2014Date of Patent: January 8, 2019Assignee: INTERNATIONAL BUSINESS MACHINES CORPORATIONInventors: Martin Oberhofer, Jens P. Seifert, Kostas Rakopoulos, Stephen Rees
-
Publication number: 20180121535Abstract: The present disclosure relates to a method for centrally processing data records using a record linkage algorithm. The method comprises providing a centralized master repository for storing data records in a predefined data structure having a set of attributes. At least one clustering metric is provided. Clusters of records may be determined using a clustering function that is based on the at least one clustering metric. For each particular cluster, a set of configuration data for the record linkage algorithm may be defined based on a value of the clustering metric within that particular cluster. The individual data records may be assigned to one or more clusters of the clusters using the clustering metric values and the record linkage algorithm may be applied to a set of two or more individual data records assigned to at least one common cluster using the set of configuration data for the common cluster.Type: ApplicationFiled: January 2, 2018Publication date: May 3, 2018Inventors: Martin Oberhofer, Yannick Saillet, Scott Schumacher, Jens P. Seifert
-
Publication number: 20180113928Abstract: The present disclosure relates to a method for centrally processing data records using a record linkage algorithm. The method comprises providing a centralized master repository for storing data records in a predefined data structure having a set of attributes. At least one clustering metric is provided. Clusters of records may be determined using a clustering function that is based on the at least one clustering metric. For each particular cluster, a set of configuration data for the record linkage algorithm may be defined based on a value of the clustering metric within that particular cluster. The individual data records may be assigned to one or more clusters of the clusters using the clustering metric values and the record linkage algorithm may be applied to a set of two or more individual data records assigned to at least one common cluster using the set of configuration data for the common cluster.Type: ApplicationFiled: October 21, 2016Publication date: April 26, 2018Inventors: Martin Oberhofer, Yannick Saillet, Scott Schumacher, Jens P. Seifert
-
Publication number: 20180096038Abstract: Embodiments of the present invention disclose generating a data profiling jobs for source data in a data processing system, the source data being described by at least one source functional data model. A target functional data model is provided, for describing target data that can be generated from the source data. One or more source functional data models are identified that correspond to the target functional data model. At least one functional source-to-target model mapping is associated to at least one source-target pair based on the target functional data model and identified source functional data models. A physical source-to-target model mapping for at least one source-target pair based on the logical source-to-target model mapping is calculated. For all physical source attributes, the needed data profiling jobs are generated based on the target attribute for analyzing the physical source attributes.Type: ApplicationFiled: December 6, 2017Publication date: April 5, 2018Inventors: Sebastian Nelke, Martin Oberhofer, Yannick Saillet, Jens P. Seifert