Aspect Term Polarity Mutual Annotation for Sentiment Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing aspect-based sentiment analysis systems fail to effectively identify and classify the polarity of different features or aspects of entities in text, particularly missing neutral and conflict polarities, and do not explicitly extract aspect terms.
Innovation Solution
A system integrating rule-based and machine learning techniques to detect and annotate aspect terms and categories, using deep syntactic parsing and mutual information to identify polarities, with a polarity inheritance mechanism from categories to undecided aspect terms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If document-level or sentence-level classification is used, then classification speed is maintained, but meaningful aspect-level sentiment data is lost
Solution Approach 1:
The patent segments the text at the aspect level by identifying aspect terms and their corresponding sentiments within sentences. Instead of classifying entire documents or sentences as a whole, the system breaks down the sentiment analysis into discrete aspect-level units, allowing simultaneous preservation of classification efficiency and extraction of granular sentiment information about specific features or aspects of entities
Solution Approach 2:
The patent introduces a new dimension of analysis by adding aspect category classification alongside traditional sentiment polarity detection. This transforms the analysis from a single-dimensional (positive/negative) classification into a multi-dimensional framework that simultaneously considers sentiment polarity and aspect category, enabling aspect-level sentiment extraction without sacrificing processing efficiency
2Device complexity
If only positive and negative opinions are considered, then classification simplicity is maintained, but neutral and conflict polarities are missed
Solution Approach 1:
The patent extends the sentiment polarity parameter from a binary state (positive/negative) to a four-state system by adding neutral and conflict polarities. This parameter expansion allows the system to capture nuanced sentiment expressions while maintaining a structured classification framework, resolving the contradiction between simplicity and comprehensiveness in sentiment categorization
3Device complexity
If aspect terms are not explicitly extracted, then processing complexity is reduced, but annotation accuracy deteriorates
Solution Approach 1:
The patent performs preliminary aspect term extraction and aspect category classification before final sentiment annotation. By identifying and labeling aspect terms and their categories in advance, the system creates a structured foundation that improves subsequent annotation accuracy while managing processing complexity through systematic pre-processing steps
4Ease of operation
If rule-based systems with handcrafted rules are used, then interpretability is maintained, but coverage of all polarity types is insufficient
Solution Approach 1:
The patent merges rule-based sentiment detection with machine learning classification components. The rule-based system provides interpretable sentiment polarity detection, while the machine learning classifier (using techniques like Naive Bayes or Maximum Entropy) handles aspect category classification and complements the rule-based approach, together achieving comprehensive coverage of all polarity types including neutral and conflict sentiments while maintaining interpretability
Data Source
AI summary
Systems and methods for aspect-based opinion mining including identifying the polarity (e.g., positive, negative, etc.) of different features of a product or a service as expressed in a text. This general task can be divided into four sub-tasks: identifying the aspect terms, classifying them into one of a set of predefined aspect categories, and identifying the polarity of the aspects terms and the aspect categories. A combination of systems (e.g., rule-based and machine learning based) can be employed to implement aspect category and aspect term polarity mutual annotation for aspect-based sentiment analysis.