Person-Centric Opinion Mining in Data Lakes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing opinion mining techniques fail to effectively aggregate and analyze disparate channel information in a multi-channel environment, providing a fragmented view of customer opinions and lacking real-time capabilities, while focusing on specific entities rather than person-centric analysis.
Innovation Solution
A method for person-centric multi-channel opinion mining that aggregates heterogeneous data from various channels into a single data repository, aligns user identities, identifies and classifies opinions, and determines sentiment polarity, enabling a comprehensive understanding of customer attitudes across multiple channels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If opinion mining techniques focus on identifying and classifying opinions on a particular target regardless of who expresses them, then entity-centric analysis is achieved, but person-centric analysis capability deteriorates
Solution Approach 1:
The patent segments the opinion mining process into distinct components: entity identification, opinion extraction, user identity extraction, and sentiment analysis. By separating the identification of opinion targets from the identification of opinion holders, the system can simultaneously perform both entity-centric and person-centric analysis without compromising either capability
Solution Approach 2:
The patent adds a new dimension to traditional opinion mining by introducing user identity extraction and alignment as a separate layer. This transforms the analysis from a single-dimension entity-focused approach to a multi-dimensional framework that incorporates both entity targets and human opiners, enabling person-centric analysis while preserving entity-centric capabilities
2Reliability
If data is kept in distinct repositories for separate channels, then channel-specific data integrity is maintained, but holistic customer view deteriorates
Solution Approach 1:
The patent merges data from multiple heterogeneous channels into a unified view by aligning user identities across channels. The system extracts user identities from each channel, aligns them to identify the same user across different channels, and aggregates their opinions, thereby creating a holistic customer view while preserving the integrity of source channel data
Solution Approach 2:
The patent introduces user identity alignment as an intermediary mechanism that connects distinct channel repositories. This intermediary layer maps users across channels without requiring direct integration of the underlying data repositories, enabling holistic analysis while maintaining channel-specific data integrity and independence
3Measurement precision
If traditional opinion mining techniques are used, then entity opinion analysis is achieved, but real-time multi-channel aggregation capability deteriorates
Solution Approach 1:
The patent performs preliminary actions by extracting and aligning user identities from multiple channels before conducting opinion analysis. By pre-processing the data to establish user identity mappings across channels, the system enables efficient real-time aggregation and analysis of multi-channel opinions without compromising analysis accuracy
Data Source
AI summary
Person-centric multi-channel opinion mining is performed in a single data repository, such as a data lake. An exemplary method comprises obtaining multi-channel heterogeneous data from a plurality of channels; identifying entities that are targets of opinion information across the plurality of channels; extracting a plurality of user identities from the plurality of channels; aligning the plurality of extracted user identities across the plurality of channels to link common user identities; identifying the entities that are targets of the opinion information of the extracted user identities; linking opinion information of the extracted user identities with a user identity associated with an opinion holder that expressed the opinion information; determining whether the opinion information comprises a positive or negative opinion; and providing a summary of the opinion information of a given opinion holder. Sentiment polarity classification algorithms optionally determine whether opinion information comprises a positive or negative opinion and assign a polarity score. An influencer score of the opinion holder is optionally assigned to the opinion information.