Multi-modal Regression for Customer Contact Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing prediction tools for customer service contact, such as regression models, fail to accurately anticipate customer needs due to their inability to effectively incorporate and process both numerical and unstructured data modalities, like text, audio, and video, which limits their predictive accuracy.
Innovation Solution
A multi-modal learning system that extracts numerical features and uses neural networks to classify customer intents from both structured and unstructured data, combining these features with regression models to estimate probabilities of customer contact, thereby enhancing prediction accuracy by incorporating all available data modalities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional regression models are used for prediction, then the system is simple to implement, but prediction accuracy deteriorates due to inability to process unstructured data
Solution Approach 1:
The patent combines traditional regression models with neural networks to create a hybrid prediction system. The neural network processes unstructured data (text, audio, video) while the regression model handles structured numerical data, and their outputs are integrated to produce the final prediction. This merging allows the system to leverage the simplicity of regression models while gaining the unstructured data processing capabilities of neural networks, thereby improving prediction accuracy without excessive complexity increase.
Solution Approach 2:
The patent creates a multi-functional prediction system that can handle multiple data modalities (structured numerical data and unstructured text, audio, video data) through a unified framework. The system uses a regression model for numerical features and neural networks for unstructured features, with both components working together to provide comprehensive customer behavior prediction. This multi-functionality allows accurate predictions across diverse data types while maintaining a cohesive system architecture.
2Measurement precision
If only numerical data is used for prediction, then processing is simple and fast, but predictive accuracy deteriorates due to lack of unstructured information
Solution Approach 1:
The patent segments the data processing into distinct pathways: one for numerical data using regression models and another for unstructured data using neural networks. Each data type is processed by the most appropriate model type, with results later integrated. This segmentation allows the system to process unstructured information (text, audio, video) without requiring the entire system to be complex, as only the relevant neural network components handle these data types while regression models continue to efficiently process numerical data.
3Measurement precision
If multi-modal data processing is implemented, then prediction accuracy improves, but computational resources and system complexity increase
Solution Approach 1:
The patent introduces an intermediary integration layer that combines the outputs of the regression model and neural network. This mediator component receives predictions from both models and integrates them into a final prediction result. The intermediary approach allows the system to process multiple data modalities accurately while managing computational complexity through a structured integration process, rather than requiring a completely complex multi-modal architecture from scratch.
Data Source
AI summary
Provided herein are systems and methods for using multi-modal regression to predict customer intent to contact a merchant. Multi-modal data including numerical data and unstructured data are extracted from customer interactions with the merchant. Features of the numerical data and the unstructured data are separately extracted and classified using techniques specific to the data types. The features for each type are then separately used to predict probabilities of customer intent. A neural network is used to combine the predictions into a single set of estimates of customer intent. This set of estimates of customer intents is used to estimate a probability that the customer will contact the merchant. The customer is then contacted based on the estimate.