Disjoint Classification Models for Sentiment Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for analyzing communications, particularly user-generated content on social media, fail to accurately determine sentiment or emotion due to their inability to analyze context, subjectivity, and polarity, leading to false positive results and making it difficult for consumers and businesses to make informed decisions.
Innovation Solution
A characteristic analytics system that uses a combination of classification models, including decision tree, neural network, and Naïve Bayes models, to parse and evaluate text objects, transcribe audio data, and determine emotions such as happiness, sadness, or sentiment as positive, negative, or neutral, by applying heuristic instructions and referencing databases to provide accurate context analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If prior methods use rules and keywords to determine sentiment, then the analysis process is simple and fast, but the accuracy is poor and false positive results occur
Solution Approach 1:
The patent segments the sentiment analysis task into multiple distinct components: context analysis, subjectivity detection, and polarity determination. Each component is handled by specialized processing modules that work sequentially, breaking down the complex analysis into manageable stages that improve accuracy without overwhelming system complexity
Solution Approach 2:
The patent adds new dimensions to the analysis by incorporating context analysis and subjectivity detection alongside traditional polarity detection. This multi-dimensional approach transforms the analysis from a single-dimensional keyword matching process into a comprehensive evaluation that considers multiple facets of the communication simultaneously
2Loss of information
If prior methods analyze only keywords, then the processing is quick and easy, but the true sentiment or emotion cannot be determined
Solution Approach 1:
The patent performs preliminary context analysis and subjectivity detection before final polarity determination. By pre-processing the communication to extract contextual information and identify subjective elements first, the system reduces information loss in subsequent analysis stages while maintaining efficient processing through staged evaluation
3Measurement precision
If multiple classification models are used to parse text, then the analysis accuracy improves, but the system complexity increases
Solution Approach 1:
The patent divides the text parsing task across multiple specialized classification models, each responsible for specific aspects such as context classification, subjectivity classification, and polarity classification. This segmentation allows each model to be optimized for its specific function, improving overall accuracy while keeping individual model complexities manageable
Solution Approach 2:
The patent designs a universal classification framework that handles multiple analysis functions (context, subjectivity, polarity) through a coordinated system of models. This multi-functional approach allows the system to achieve comprehensive analysis accuracy while reusing common processing infrastructure across different classification tasks
Data Source
AI summary
Systems and methods for providing characteristic analytics are provided. A data construct including a text object is received. A first subset of classification models in a plurality of classification models is applied to the data construct. The subset of classification models parse the text object into a plurality of text strings. Each respective classification model in the plurality of classification models parses a portion of the text object in accordance with a plurality of heuristic instructions associated with the respective classification model. The text object is evaluated using a reference database that includes a predetermined plurality of text strings. A characteristic of the data construct in the form of a result of the evaluating is provided. The evaluation step further comprises a second disjoint subset of classification models used for the evaluation of the parsed text object.