Microsoft Technology Licensing, LLC

Jason Shaver

Neelakantan Sundaresan

Roshanak Zilouchian Moghaddam

Spandan Garg

Kesavan Shanmugam

Jared Porter Parsons

Tomas Matousek

Yevhen Mohylevskyy

Jobst-Immo Landwerth

Matthew Glenn Jin

Gearard Boland

Anshul Gupta

Gabriel Pedro De Castro

Payal Bajaj

Richard Steele Gibson

Yuval Mazor

Eduardo Enrique Noriega De Armas

Claire Novotny

Software provenance validation reports whether a validation binary matches the source code, resources, and other parts, as well as the compiler, runtime, operating system, and other context, which is specified in a provenance manifest for a release binary. Part context checksums, software versions, tool parameters, and other aspects of a build are checked. Certification signatures, timestamps, certain version differences, source code locations, and other data may be ignored for validation purposes. A provenance manifest may include other provenance manifests, including binary rewrite manifests. The provenance manifest may be stored in a debugger file with symbol information, or stored separately. Partial matches may be reported, with details of what matches or does not match. After provenance of a binary is validated, the binary's source code can be analyzed for vulnerabilities, thereby enhancing software supply chain security.

Software provenance validation

Detecting whether or not an open source software package has functionality which is not described by the source code used to build the open source software package. To do so, in one embodiment, this is done by accessing source code used to build the open source software package. The open source software package is built from the source code. After the open source software package has been rebuilt, then it is computed whether or not the rebuilt package accomplishes the same functions as the open source software package. Finally, if the rebuilt package does not accomplish the same functions as the open source software package, an alert is raised.

Detecting added functionality in open source package

Examples of the usage of a command of a command line interface includes the command with a set of parameters and corresponding parameter values. The examples are generated from telemetry data, which does not contain parameter values, and from web-based sources that may contain multiple parameter values. A machine learning model is used to predict the data type of a parameter value when the parameter is used with a particular command. The predicted data type is then used to select an appropriate parameter value for the example from multiple known parameter values or to generate a parameter value when no known parameter value exists.

Machine-generated examples of command-line commands with parameter values

An automated command assistance tool is provided for a browser-enabled command line interface of a cloud service. The automated command assistance tool provides examples illustrating the correct syntax for commands used to manage the resources of a cloud service. The command assistance tool learns the syntax of a command from usage patterns found in telemetric data, scripts and user documentation and forms templates containing a command's usage pattern and related information. The templates are used to generate examples that respond to a user query for assistance with usage of a particular command.

Automated command assistance for a cloud service

A term-weighting and document-scoring function is used to search for a command line interface (CLI) script that is likely relevant to an operation specified in a natural language query. CLI scripts are created to perform various operations of a CLI-based application. A CLI script is associated with a description document having keywords associated with the individual commands used in the CLI script. The relevance of a CLI script to an intended operation is based on the term-weighting and document-scoring function which is applied to each component of each command in a CLI script and weighted accordingly.

Command-line script generation with relevance search

An interactive question and answer (Q&A) service provides pairs of questions and corresponding answers related to the content of a web page. The service includes pre-configured Q&A pairs derived from a deep learning framework that includes a series of neural networks trained through joint and transfer learning to generate questions for a given text passage. In addition, pre-configured Q&A pairs are generated from historical web access patterns and sources related to the content of the web page.

Question and answer pair generation using machine learning

Growing community of inventors

Redmond, WA, United States of America

Jason Shaver