Co-occurrence Graph Visualization for Sentiment Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is a need for an automated system to process and summarize vast amounts of opinion data from social media and other online sources to provide actionable insights for brand managers and marketers regarding consumer preferences and sentiment towards specific brands and their competitors.
Innovation Solution
The system employs graphical representations, such as Instance Graphs and Co-occurrence Graphs, to visualize frame instance data and co-occurrence data, using frame extraction rules and linguistic analysis to identify preferences and relationships between items, allowing for the creation of Preference Frames and the generation of graphical plots that illustrate net preferences and influence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If vast amounts of opinion data are collected from social media and online sources, then the quantity of available information increases, but the complexity of processing and summarizing this data increases
Solution Approach 1:
The patent segments the vast opinion data into structured frame instances based on predefined frames (e.g., Preference Frame, Entity Frame). Each frame instance represents a discrete unit of structured information extracted from unstructured text, enabling systematic processing and analysis of large volumes of data without overwhelming system complexity
Solution Approach 2:
The patent introduces frame-based structured data as an intermediary layer between raw opinion data and analytical insights. Frame instances serve as standardized intermediaries that transform unstructured social media text into organized, queryable data structures, simplifying subsequent processing and analysis operations
2Measurement precision
If frame extraction rules and linguistic analysis are applied to identify preferences and relationships, then the precision of sentiment analysis improves, but the processing time and computational resources increase
Solution Approach 1:
The patent applies frame extraction rules and linguistic analysis in advance to pre-process and structure opinion data into frame instances before detailed sentiment analysis. This preliminary structuring organizes data into standardized formats with identified entities, preferences, and relationships, reducing the computational burden during subsequent analysis phases
Solution Approach 2:
The patent transforms unstructured text data into structured frame instances by changing the parameter representation from free-form text to standardized fields (e.g., entity names, preference directions, relationship types). This parameter transformation enables more efficient processing and querying while maintaining high precision in sentiment identification
3Ease of operation
If graphical representations are created to visualize frame instance data and co-occurrence data, then the ease of understanding consumer preferences improves, but the device complexity for data visualization increases
Solution Approach 1:
The patent transforms structured frame instance data into graphical visualizations by adding spatial dimensions. Co-occurrence graphs and preference visualizations map abstract data relationships onto two-dimensional visual spaces, enabling intuitive understanding of consumer preferences and sentiment patterns through visual positioning, connections, and layouts
Data Source
AI summary
Occurrence and co-occurrence data can be presented as a graph. A Non-directional Co-occurrence Graph represents occurrences of an item x as a node and co-occurrences, of items x and y, as an edge between the nodes for x and y. The size of a node can depend on the log, or a scaled value, of its occurrences. A Directional Co-occurrence Graph is the same as a Non-directional Co-occurrence Graph, except there can be, between each pair of nodes, up to two directed edges. An edge pointing from x to y can be used to represent the following ratio: number of co-occurrences of x and y divided by the number of occurrences of x. More specifically, the thickness of the edge can be proportional to the ratio. Net sentiment can be determined for an item x and used to modify the graphical representation (e.g., the color) of a node or edge.