Sentiment Analysis in Interactive Response Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Interactive response systems, such as IVR and chatbots, often generate suboptimal responses that frustrate users, leading to negative experiences and lost business opportunities, as they fail to identify and address user sentiment in real-time due to the complexity and resource-intensive nature of sentiment analysis using machine learning.
Innovation Solution
The system preprocesses user utterances using a natural language processor, signal extractor, and utterance sentiment classifier to reduce data input for a sequential neural network classifier, enabling faster and more efficient sentiment scoring and response selection, thereby allowing for real-time user sentiment analysis and improved response generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning is used to predict sentiment of a conversation, then user sentiment can be identified, but the volume of training data required is huge and considerable processing resources are needed resulting in delayed output
Solution Approach 1:
The patent segments the conversation into discrete utterances and processes each utterance independently through the ML algorithm. This segmentation allows the system to analyze sentiment in smaller, manageable units rather than processing entire conversations at once, reducing the computational burden and data input requirements while maintaining accurate sentiment identification for each utterance in context
Solution Approach 2:
The system performs preliminary processing by maintaining a context window of previous utterances and pre-processing this data before feeding it to the ML algorithm. This preliminary action prepares the data in advance, allowing the ML model to focus on sentiment analysis rather than raw data processing, thereby reducing real-time computational requirements while preserving contextual accuracy
2Measurement precision
If machine learning is used to predict sentiment of a conversation, then user sentiment can be identified, but delayed output occurs due to considerable processing resources required
Solution Approach 1:
By segmenting the conversation into individual utterances and processing them sequentially with a fixed context window, the system enables faster, incremental sentiment analysis. Each utterance is processed independently with previously processed utterances providing context, allowing real-time sentiment identification without waiting for complete conversation analysis
Solution Approach 2:
The system uses a fixed-size context window that includes only the necessary number of previous utterances (e.g., last 3-5 utterances) rather than analyzing the entire conversation history. This partial action approach provides sufficient contextual information for accurate sentiment identification while significantly reducing processing time and computational resources compared to analyzing all previous utterances
3Loss of information
If a large volume of data is input into the ML to understand the utterance in context, then contextual sentiment can be understood, but processing resources and time increase considerably
Solution Approach 1:
The patent divides the contextual information into discrete, manageable utterance units and processes them in a sequential manner. Each utterance is analyzed with a fixed window of previous utterances providing context, enabling efficient processing while maintaining contextual understanding through structured segmentation of the conversation data
Solution Approach 2:
The system changes the parameter of context window size to a fixed, optimized value that balances contextual understanding with processing efficiency. By adjusting this parameter, the system can maintain sufficient contextual information for accurate sentiment analysis while preventing excessive data input that would overwhelm processing resources and delay output
Data Source
AI summary
Apparatus and methods for leveraging machine learning and artificial intelligence to generate a response to an utterance expressed by a user during an interaction between an interactive response system and the user is provided. The methods may include a natural language processor processing the utterance to output an utterance intent. The methods may also include a signal extractor processing the utterance, the utterance intent and previous utterance data to output utterance signals. The methods may additionally include an utterance sentiment classifier using a hierarchy of rules to extract, from a database, a label, the extracting being based on the utterance signals. The methods may further include a sequential neural network classifier using a trained algorithm to process the label and a sequence of historical labels to output a sentiment score. The methods may further include, based on the utterance intent, the label and the score, to output a response.