Microsoft Technology Licensing, LLC

Jeffrey Bernhardt

Bekim Demiroski

David Noor

Matthew Carroll

David Maxwell Chickering

Robert L Rounthwaite

C James MacLennan

Pyungchul Kim

David Earl Heckerman

Christopher A Meek

Dmitry Sonkin

Dax H Hawkins

Ciprian Gerea

James Kelsey Howey

Bruce A Prang

Theodore Tak Lee

James Kirk Haselden

Ilya Vinarsky

A system and method for facilitating execution of one or more data integration projects in multiple environments or an environment that undergoes changes. Each project has a set of project parameters, which are bound to environment variables. Each environment has a corresponding environment representation with environment variables and corresponding values. Each project is mapped to an environment representation. Values of environment variables are provided to projects with corresponding parameters. When one or more projects are changed to a different environment with a different corresponding environment representation, the environment variable values of the new environment representation are provided to the projects. When an environment change is reflected in the environment representation, the changed variable values are provided to mapped projects.

Multi-environment configuration of data integration projects

A predictive model is used to populate a cache in a videogame system. The predictive model takes as an input a sequence of file sections that have been requested for the associated videogame thus far. The predictive model then returns the names or indicators of one or more file sections that will likely be requested in the future along with a probability that those file sections will be requested. A cache driver executing in the background of the videogame system may then use the predictive model to reduced load times during the execution of a videogame by retrieving the predicted file sections from the optical disk and placing them in a cache during periods of low CPU activity.

Predictive caching of assets to improve level load time on a game console

A script driven infrastructure is provided in order to selectively move structure and data from a source database to a target database. The structure and data is moved using a data transfer mechanism, where the data is contained in disk files stored on the source database and it has corresponding script files describing the structure of the data. The transfer of the data is driven by the script files which allows for selective transfer of structure and data. In one aspect of the invention, the disk files are stored in SQL format and the script files are saved in SQL format and XML format, where the script files saved in SQL format create objects and relational constraints in the target database, and the script files stored in XML format drive the transfer process of the data stored in the disk files.

System and method for transferring data and metadata between relational databases

The operation of a data flow is improved by providing for validation of the consistency of the data flow with the external data files it works with, even in the face of changes in those external data files Metadata is stored regarding the external data files. Because external data files may change, this metadata is used when determining whether the external data files are still compatible with the data flow. In such performing such validation, the metadata tracks what changes in the external data files have occurred, and allows increased flexibility in presenting the user with options or in automatically fixing the data flow to correspond with changes in external data files. When the external data files are not available, the stored metadata can be used to validate the data flow to the last information regarding the external data files. The external metadata can be resynchronized to the data flow or to the external data files to provide updated external metadata for later validation.

External metadata processing

Selection of certain attributes as output and input attributes is provided so a decision tree may be created more efficiently. For each possible output attribute an interestingness score is calculated. This interestingness score is based on entropy of the output attribute and a desirable entropy constant. The attributes with the highest interestingness score are used as output attributes in the creation of the decision tree. Score gains for the input attribute over the output attributes are calculated using a conventional scoring algorithm. The sum of the score gains over all output attributes for each input attribute is calculated. The attributes with the highest score gain sums are used as input attributes in the creation of the decision tree.

System and method for feature selection in decision trees

Continuous attributes are used as input attributes in decision tree creation. Buckets are created by dividing the range of values for the continuous attribute into sub-ranges of equal extent. These buckets form initial partitions. Mergers of adjacent partitions are considered to determine score gains from such mergers, and the most useful mergers occur. The resulting partitions are used as the discretization of the continuous attribute for use as an input attribute.

System and method for handling a continuous attribute in decision trees

A system and method for use with a data mining application for a large database having a large number of records. A selection attribute is chosen from one of a plurality of attributes contained by records within the database. Records are scanned in the database and a randomizing function is applied to the selection attribute of each record to create a randomized record value. A selection criteria is then applied to identify records for inclusion within a subset of records (smaller than the original data set) by comparing the randomized record value of each record with the selection criteria. The subset of records having a randomized record value satisfying the selection criteria approximates the entire database but takes up less memory and can be evaluated or scanned much more quickly.

Server side sampling of databases

Disclosed is a system for approximating conditional probabilities using an annotated decision tree where predictor values that did not exist in training data for the system are tracked, stored, and referenced to determine if statistical aggregation should be invoked. Further disclosed is a system for storing statistics for deriving a non-leaf probability corresponding to predictor values, and a system for aggregating such statistics to approximate conditional probabilities.

System and method for approximating probabilities using a decision tree

Determination as to whether a variable is numeric or non-numeric. In one embodiment, a variable is input having a plurality of values, where each value has a count. The variable is determined to be numeric or non-numeric by assessing closeness of counts for adjacent values of the variable. Whether the variable is numeric or non-numeric is then output.

Growing community of inventors

Woodinville, WA, United States of America

Jeffrey Bernhardt