Topical Sentiment Analysis Using Segmented Text Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current text mining techniques fail to effectively combine polarity detection with topicality, leading to inefficiencies in identifying consumer sentiments towards specific products within large volumes of unstructured data, as they either overlook topical relevance or rely on assumptions that do not generalize across domains.
Innovation Solution
A computer-implemented method that determines the topic and polarity of document segments by using a domain-general polarity lexicon, syntactic, and semantic rules, along with machine learning classifiers to identify and associate polar expressions with their respective topics, enabling the extraction of topical sentiments from electronically stored communications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a whole-document approach is used for sentiment detection, then the preponderance of expressed sentiment can be captured, but the relationship between polar language and topicality is overlooked
Solution Approach 1:
The patent segments documents into sentences or phrases for individual analysis, allowing both sentiment detection and topicality assessment at the sentence level. This segmentation enables the system to identify polar expressions while simultaneously determining their topical relevance, resolving the contradiction between processing efficiency and measurement precision.
Solution Approach 2:
The patent adds a topicality dimension to traditional sentiment analysis by classifying sentences into topic categories alongside polarity classification. This dimensional expansion allows the system to maintain high productivity through automated classification while achieving precise measurement of topical relevance through multi-label classification.
2Measurement precision
If a fine-grained NLP-based textual analysis is used to capture local context, then the subject of polar expression can be identified, but the topic is captured by vague base noun words
Solution Approach 1:
The patent employs a unified classification framework that simultaneously performs polarity detection and topic classification using the same sentence-level input. This multi-functional approach allows the system to accurately identify polar expressions while also determining their topics, eliminating the limitation of using only vague base noun words for topic identification.
Solution Approach 2:
The patent introduces topic classification as an intermediary step between sentiment detection and topic identification. By classifying sentences into topics as an intermediate process, the system can accurately identify polar expressions while also determining their specific topics, resolving the contradiction between measurement precision and adaptability.
3Measurement precision
If manual review of customer service emails is performed, then accurate understanding of public sentiment can be achieved, but the process is time-consuming and inefficient
Solution Approach 1:
The patent implements automated text classification systems that perform sentiment and topic analysis independently without human intervention. The system self-services by automatically processing large volumes of emails, identifying polar expressions, and classifying topics, thereby achieving high measurement precision while eliminating the time loss associated with manual review.
Solution Approach 2:
The patent replaces manual mechanical review processes with automated computational classification systems. By substituting human analysis with machine learning-based classification algorithms, the system achieves both high accuracy in sentiment understanding and significant time savings through rapid automated processing.
Data Source
AI summary
The present application presents methods for performing topical sentiment analysis on electronically stored communications employing fusion of polarity and topicality. The present application also provides methods for utilizing shallow NLP techniques to determine the polarity of an expression. The present application also provides a method for tuning a domain-specific polarity lexicon for use in the polarity determination. The present application also provides methods for computing a numeric metric of the aggregate opinion about some topic expressed in a set of expressions.