18049598. MEASURING PROBABILITY OF INFLUENCE USING MULTI-DIMENSIONAL STATISTICS ON DEEP LEARNING EMBEDDINGS simplified abstract (Microsoft Technology Licensing, LLC)

MEASURING PROBABILITY OF INFLUENCE USING MULTI-DIMENSIONAL STATISTICS ON DEEP LEARNING EMBEDDINGS

Organization Name

Microsoft Technology Licensing, LLC

Inventor(s)

Amund Tveit of Trondheim (NO)

Kateryna Solonko of Oslo (NO)

Zoran Hranj of Oslo (NO)

Aleksander Werpachowski of Oslo (NO)

Roman Werpachowski of Ameberg (NO)

Mohammadreza Bonyadi of Trondheim (NO)

Arun Shankar Iyer of Bangalore (IN)

MEASURING PROBABILITY OF INFLUENCE USING MULTI-DIMENSIONAL STATISTICS ON DEEP LEARNING EMBEDDINGS - A simplified explanation of the abstract

This abstract first appeared for US patent application 18049598 titled 'MEASURING PROBABILITY OF INFLUENCE USING MULTI-DIMENSIONAL STATISTICS ON DEEP LEARNING EMBEDDINGS

Simplified Explanation

The disclosure describes a system for measuring the probability of influence in digital communications to determine if communication content originated in a person's own prior knowledge or new information obtained from interactions with others.

• The system uses multidimensional statistics on embeddings representing communications to estimate the probability that a new communication by a first user comes from the same distribution as their prior communications.
• It also generates an estimated probability that the new communication comes from the same distribution as communications of a second user that were accessible to the first user.
• If the second probability is greater than the first probability, it suggests that the new communication is more likely influenced by exposure to the second user's communications rather than the first user's own historical knowledge.
• An influence attribution recommendation is then generated, including an influence attribution or other recommended action.

Potential Applications

This technology could be applied in social media platforms to identify the source of information shared by users and to detect potential instances of influence between users.

Problems Solved

This technology helps in determining the origin of communication content, distinguishing between a user's own knowledge and information obtained from interactions with others. It can help in understanding the influence dynamics in digital communications.

Benefits

- Improved understanding of information sharing dynamics - Enhanced ability to attribute influence in digital communications

Potential Commercial Applications

A potential commercial application of this technology could be in social media analytics tools that provide insights into user interactions and influence dynamics on the platform.

Possible Prior Art

One possible prior art could be research on social network analysis and influence detection algorithms used in digital marketing to track the impact of influencers on consumer behavior.

How does this technology handle privacy concerns related to analyzing user interactions in digital communications?

The system should have built-in privacy measures to ensure that user data is anonymized and protected during the analysis process. Additionally, user consent and data protection regulations should be considered in implementing this technology.

Can this technology be adapted for real-time analysis of communication content to provide immediate influence attribution recommendations?

The system may need to be optimized for real-time processing of communication data to provide timely influence attribution recommendations. This could involve enhancing the efficiency of the statistical analysis algorithms and data processing capabilities.

Original Abstract Submitted

The disclosure herein describes a system for measuring probability of influence in digital communications to determine if communication content originated in a person's own prior knowledge or new information more recently obtained from interaction with communications of others. An estimated probability a new communication by a first user comes from the same distribution as prior communications of the first user are generated using multidimensional statistics on embeddings representing the communications. A second estimated probability that the new communication comes from the same distribution as communication(s) of a second user that were accessible to the first user are generated. If the second probability is greater than the first probability, the new communication is more likely influenced by exposure of the first user to the second user's communications rather than the first user's own historical knowledge. An influence attribution recommendation is generated, including an influence attribution or other recommended action.