Conversation Sentiment Scoring Using Rule and Neural Network Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Interactive response systems, such as IVR and chatbots, struggle to identify and respond to user sentiment in real-time, leading to suboptimal responses that can frustrate users and result in negative reviews, as existing sentiment prediction methods require large training data sets and significant processing resources, making them inefficient and delayed.
Innovation Solution
The implementation of a conversation sentiment scoring system using a conversation manager, natural language processor, signal extractor, utterance sentiment classifier, and sequential neural network classifier, which processes user utterances to generate sentiment scores and labels, allowing the system to select appropriate responses and reduce resource consumption by using pre-processing and training with label data only.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If machine learning is used to predict sentiment of a conversation, then sentiment prediction capability is improved, but processing resources required and output delay increase
Solution Approach 1:
The patent segments the conversation into discrete utterances and processes each utterance independently through the ML algorithm. Instead of analyzing the entire conversation at once, the system breaks it down into smaller units that can be processed sequentially, reducing the computational burden and improving processing speed while maintaining sentiment prediction accuracy.
Solution Approach 2:
The patent implements pre-processing steps that extract relevant features from utterances before feeding them to the ML algorithm. This preliminary action prepares the data in advance, allowing the ML model to work with pre-processed features rather than raw text, thereby reducing processing time and resource requirements during actual sentiment prediction.
2Measurement precision
If machine learning is used to predict sentiment with context understanding, then sentiment accuracy is improved, but data input volume and processing resources increase
Solution Approach 1:
The patent extracts only the necessary contextual information from previous utterances and relevant features from current utterances, rather than processing the entire conversation history. This extraction approach maintains sentiment accuracy by including only the most relevant data points while significantly reducing the volume of data that needs to be input to the ML algorithm.
Solution Approach 2:
The patent applies different processing approaches to different parts of the conversation. Previous utterances are processed to extract only sentiment-relevant features, while current utterances receive full text analysis. This local quality approach ensures that each part of the input receives the appropriate level of processing, maintaining accuracy while reducing overall data volume.
3Measurement precision
If large training data sets are used to train the ML algorithm, then sentiment recognition accuracy is improved, but training complexity and time increase
Solution Approach 1:
The patent changes the parameters of the training data by focusing on labeled sentiment data rather than requiring large volumes of raw conversational data. By transforming the training approach to use labeled datasets with sentiment annotations, the system achieves high recognition accuracy with reduced training complexity and shorter training time.
Data Source
AI summary
Apparatus and methods for leveraging machine learning and artificial intelligence to assess a sentiment of 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.