Sentiment Analysis for Microblogs Using Lexicon Rules
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Sentiment analysis on microblogs is challenging due to their short length, noisy data, and domain-specific expressions, which complicate the identification of opinion words and their polarities, especially in interrogative sentences and when multiple words with opposite polarities appear together.
Innovation Solution
A system employing data cleansing techniques and a rules-driven opinion lexicon for sentiment analysis, including pre-processing to filter out spam and expand abbreviations, part-of-speech tagging, and a sentiment polarity assignment engine that accounts for domain-specific words and context, resolves opinion polarities by using a domain-specific lexicon builder and sentiment analyzer.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data cleansing and pre-processing are applied to filter noise and expand abbreviations, then sentiment analysis accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent applies data cleansing and pre-processing techniques before sentiment analysis to filter noise, remove spam, and expand abbreviations. This preliminary action prepares the microblog data in advance, improving the accuracy of subsequent sentiment analysis while managing computational complexity through structured preprocessing steps.
2Measurement precision
If a rules-driven opinion lexicon with domain-specific words is used, then identification of opinion words in context is improved, but processing time increases
Solution Approach 1:
The patent uses a rules-driven opinion lexicon that incorporates domain-specific words and contextual parameters. By changing the parameters of the lexicon to include domain-specific terminology and contextual rules, the system improves opinion word identification accuracy while managing processing time through efficient rule-based matching.
3Adaptability or versatility
If microblog characteristics including emoticons and colloquial expressions are analyzed, then sentiment detection capability is improved, but computational complexity increases
Solution Approach 1:
The patent addresses microblog characteristics by applying local quality principles to handle different data types. The system processes emoticons, colloquial expressions, and domain-specific language with specialized rules and lexicon entries tailored to each characteristic, improving sentiment detection capability while managing computational complexity through targeted processing approaches.
Data Source
AI summary
There is provided a computer-implemented method of performing sentiment analysis. An exemplary method comprises identifying one or more sentences in a microblog. The microblog comprises an entity. The method further includes identifying one or more opinion words in the sentences based on an opinion lexicon. Additionally, the method includes determining, for each of the sentences, an opinion value for the entity. The opinion value is determined based on an opinion value for each of the opinion words in an opinion lexicon.