Microsoft Technology Licensing, LLC

Electrifai Opco, LLC

LuMing Wang

Xiaohong Mark Yang

Patrick John Baumgartner

Peter Eberhardy

Mark Xiaohong Yang

Zhenyu Tang

Alisson A S Sol

Sonal Jain

Hailei Zhang

Jin Wang

Malik Hussain

Zaiwei Du

Amer Shahnawaz

Li Deng

Jianfeng Gao

Xiaodong He

Zijian Zheng

Jianshu Chen

Geoffrey Xiao

Pablo Jose Barvo

Prabhdeep Singh

Anton Amirov

Mingyang Zhao

Alisson Sol

Jason Malcolm

Yilei Zhang

Brendan W L Clement

Bochen Jin

Sandeep P Solanki

Shahin Rahman

Hanjun Xian

Jinhua Ma

A method comprising receiving data associated with a business, the data comprising first values for first attributes; processing the data, in accordance with a common data attribute schema that indicates second attributes, to generate second values for at least some of the second attributes including a group of attributes, the second values including a group of attribute values for the group of attributes; identifying, using the common data attribute schema and from among pre-existing software codes, software code implementing an ML data processing pipeline configured to generate a group of feature values; processing the group of attribute values with the software code to obtain the group of feature values; and either providing the group of feature values as inputs to a machine learning (ML) model for generating corresponding ML model outputs, or using the group of feature values to train the ML model.

Systems and methods for generating and deploying machine learning applications

A processing unit can acquire datasets from respective data sources, each having a respective unique data domain. The processing unit can determine values of a plurality of features based on the plurality of datasets. The processing unit can modify input-specific parameters or history parameters of a computational model based on the values of the features. In some examples, the processing unit can determine an estimated value of a target feature based at least in part on the modified computational model and values of one or more reference features. In some examples, the computational model can include neural networks for several input sets. An output layer of at least one of the neural networks can be connected to the respective hidden layer(s) of one or more other(s) of the neural networks. In some examples, the neural networks can be operated to provide transformed feature value(s) for respective times.

Training and operation of computational models

An object oriented search mechanism extracts structural metadata and data based on type of document contents and data sources connected to the documents. Relationships between textual and non-textual elements within documents as well as metadata associated with the elements and data sources are utilized to generate a unified object model with the addition of semantic information derived from metadata and taxonomy, which are used to enhance search indexing, ranking of search results, and dynamic adjustment of result rendering user interface with fine tuned relevancy. Additional data from data sources connected to the documents may also be used to unlock hidden data such as data that has been filtered out in an original document.

Object oriented data and metadata based search

Technologies are described herein for defining the scope of a sub-cube based upon a database query. A request is received to perform a query of a multi-dimensional database, such as a cube, stored at a server computer. In response to the request, the scope for a sub-portion, called a sub-cube, of the multi-dimensional database is determined based upon the query. The scope of the sub-cube may be determined by identifying the minimum sub-portion of the cube necessary to perform the query. Once the scope has been determined, the sub-cube is generated at the server computer and transmitted to a client computer. The client computer stores the sub-cube locally and performs the requested database query on the sub-cube stored at the client computer. A user interface may be provided through which a user can modify the scope of the sub-cube determined based upon the query.

Defining sub-cube scope based upon a query

Relationships between linked and/or embedded documents as well as documents sharing data source(s) are captured and rendered through virtual documents. Virtual documents are created representing linked/embedded documents and data sources associated with a relevant document. Relationships between real and virtual documents are preserved and rendered along with search results providing a user a comprehensive picture of search results.

Indexing and searching employing virtual documents

Data processing architecture where submitters can write data and/or metadata changes to a destination through an asynchronous interface. The architecture includes a ticketing system that issues a ticket (a unique identifier) to the submitter in response to receiving a changelist from the submitter. When the changes are successfully completed at the destination, the cached destination data can be re-downloaded to the submitter, and the changes cached in the submitter, discarded. Absent this capability, a user has to manually manage the cached changes and cached data. The architecture can be distributed in that multiple submitters can submit changelists. The destination provides data integrity by managing submissions to assist in ensuring data integrity across multiple submissions. Automatic changelist cleanup is provided on the submitter so that changelists do not impact performance, memory, and storage.

Maintaining client data integrity in a distributed environment using asynchronous data submission

An interface for designing a report is provided, with elements of the report being disposed at intersections of report positions and configured to present information derived from data in a database. A formula is associated with one or more selected elements in the report to generate a result for the one or more selected elements. A tuple identifying the report positions of the one or more selected elements is associated and stored with the formula in a formula list associated with the report. The report is configured so that, when the report is generated, it is determined whether identifying report positions corresponding to an identified element in the report correspond with the tuple in the formula list. Upon determining the tuple corresponds with the identifying report positions for the identified element, the formula associated with the tuple is applied to generate the result.

Method for capturing design-time and run-time formulas associated with a cell

Techniques to manage information for dynamic reports are described. An apparatus may comprise a business planning application module to update a current period reference to a time member of a time dimension, generate a rolling data entry form with actual data members and forecast data members based on the current period reference, and set a data locking mode for the actual data members and a data access mode for the forecast data members. Other embodiments are described and claimed.

Techniques to manage information for dynamic reports using time periods for a business model

Tools and techniques are described for selecting member sets for generating asymmetric queries. User interfaces provided by this description may include representations of different dimensions that include respective members. These dimensions define hierarchical data structures against which queries are run to generate requested reports. The user interfaces may include representations of members associated with different dimensions, with members from different dimensions arranged in selected orders. The user interfaces may also provide selection tools that activate symmetrical or asymmetrical rendering modes for constructing the query. In the symmetrical rendering mode, the query cross-joins all of the members selected from one dimension with all of the members selected from the other dimension. In the asymmetrical rendering mode, the query cross-joins the first-ordered member from one dimension with the first-ordered member from another dimension, cross-joins the second member from one dimension with the second member from another dimension, and so on.

Selecting member sets for generating asymmetric queries

A computer-implemented method includes selecting a first member of a first dimension of a database to be included at a first location in a report. The first member includes first data to be retrieved from the database. A blank space to be added at a second location relative to the first member is also selected. A database query is generated that includes a first instruction to retrieve first data included in the first member. The first instruction is located at a first position in the query and the first data will appear in the report at the first location. The database query also includes a second instruction to retrieve a null set of data. The second instruction is located at a second position in the query and the null set of data will appear in the report at the second location.

Rendering database queries with blank space

Growing community of inventors

Bellevue, WA, United States of America

LuMing Wang