Perception Capture System for Brand Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Measuring brand perception is challenging due to biases in traditional customer survey methods and limitations in data scale, and existing approaches to analyze social media data often miss important concepts relevant to a brand as they rely on count-based methods that fail to capture concepts mentioned infrequently.
Innovation Solution
A perception capture system using deep learning to analyze social media data, employing word embeddings and similarity scoring methods like cosine similarity to generate perception scores, which are stable across time periods and robust to noise, allowing for the identification of top concepts representing brand perception and monitoring changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional customer survey methods are used to measure brand perception, then the measurement process is simple to implement, but the results suffer from sampling bias and limited data scale
Solution Approach 1:
The patent replaces traditional mechanical survey methods with an automated deep learning system that processes social media data. The system uses neural networks to generate word embeddings and calculate perception scores, substituting manual survey processes with computational algorithms that analyze large-scale social media mentions automatically.
Solution Approach 2:
The patent introduces social media data as an intermediary between the brand and the measurement system. Instead of directly surveying customers, the system analyzes public social media mentions, using this intermediate data source to infer brand perception while avoiding the biases inherent in direct survey methods.
2Measurement precision
If count-based methods are used to analyze social media data, then the analysis process is computationally simple, but important concepts mentioned infrequently are missed
Solution Approach 1:
The patent transforms the analysis from counting word frequencies to calculating semantic similarity scores using cosine similarity. This parameter change allows the system to identify important concepts based on their semantic relationship to the brand rather than their frequency, capturing infrequently mentioned but highly relevant concepts.
Solution Approach 2:
The patent moves the analysis from a one-dimensional frequency count to a multi-dimensional semantic space using word embeddings. By representing words as vectors in high-dimensional space, the system can detect concepts based on their semantic proximity to the brand, adding a new dimension of analysis that captures meaning beyond frequency.
3Reliability
If perception scores are calculated using deep learning models, then the scores are stable and robust to noise, but the computational complexity and processing time increase
Solution Approach 1:
The patent performs preliminary action by pre-training deep learning models to generate word embeddings before the actual perception scoring. This pre-processing step creates a reusable semantic space that can be applied to multiple brands and time periods, reducing the computational burden during actual measurement while maintaining score stability and robustness.
Data Source
AI summary
Systems and methods are provided to generate a set of vector representations for each word of a plurality of words in a set of social media data by inputting each word into each of a predefined number of machine learning models to output each of the set of vector representations, the set of social media data comprising social media data from a plurality of social network platforms in a predefined window of time. The systems and methods further provide for generating a similarity score for each vector representation in the set of vector representations for each word output from the predefined number of machine learning models with respect to a given brand name, and generating a perception score for each word based on the generated similarity score for each vector representation in the set of vector representations for each word.