International Business Machines Corporation

Maciej Madej

Tymoteusz Gedliczka

Szymon M Brandys

Tomasz P Kania

Krzysztof Pitula

Piotr Grzywna

A computer-implemented method, a computer program product, and a computer system for measuring relevance of datasets to data science models. One or more servers implement steps: extract keywords in each data science model; determine first relative frequencies of the respective keywords in each data science model, for each source group in the data science models; extract keywords in each dataset; determine second relative frequencies of the respective keywords in each dataset, for each source group in the datasets; determine weights of the keywords; calculate first aggregated relevant scores of the respective keywords in each data science model, based on the first relative frequencies and the weights; calculate second aggregated relevant scores of the respective keywords in each dataset, based on the second relative frequencies and the weights. One or more servers calculate similarity between vectors of the first and second aggregated relevant scores, based on a similarity measure between vectors.

Measuring relevance of datasets to a data science model

A computer-implemented method for quick data exploration of data to be uploaded may be provided. The method comprises uploading, from a local system, a first data set of a first size and determining that the first data set is not corrupted. The method also comprises in parallel to the uploading performing selecting from the first data set a predefined number of records and building a second data set, determining statistical data and metadata about the first data set, and visualizing the second data set, the statistical data and the metadata.

Quick data exploration

Technology for making corrections to URL and/or URI requests (herein generically referred to as 'UR requests') based on machine learning of patterns that can be gleaned from previously received UR requests. The patterns are then used to correct new UR requests that are received with errors in the resource location path. In some embodiments, the errors detected and corrected are path errors. In some embodiments the path error is an erroneous inclusion of an extraneous reserved slash ('/') character, which, when detected, is replaced in the UR string with its corresponding proper code (for example, - -%2F- -).

Correcting a URL within a REST API call

A computer-implemented method that can prevent an upload of a data set upon detection of a modification of the data set. The method includes storing a first portion of a file in a buffer while being in a receiving mode. Upon determining that applying one of a predefined transformation performed to the first portion of an existing data set will result in reproducing a content of the buffer: generating a transformation notification signal; and upon receiving a stop message, stopping the receiving mode which results in using the existing file on the server.

Preventing unnecessary upload

A computer-implemented method for loading data into a secure storage volume may be provided. The method comprises receiving data to be stored on the secure storage volume, storing the received data in a buffer, and upon determining that the secure storage volume is unlocked, transferring the received data from the buffer to the secure storage volume in encrypted form.

Growing community of inventors

Cracow, Poland

Maciej Madej