Next-Step Sentence Prediction for Communication Session Analytics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital communication platforms lack the ability to provide intelligent prediction and analytics for next step sentences in communication sessions, particularly in sales meetings, which are crucial for measuring and improving sales team performance.
Innovation Solution
A system that defines annotation guidelines, trains a model using labeled training data, and employs ensemble class scores to predict next step sentences through a combination of language models and classification heads, providing actionable insights to users.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If traditional communication platforms are used, then basic communication functionality is provided, but intelligent prediction and analytics for next step sentences are not available
Solution Approach 1:
The patent embeds multiple AI components within each other: language models are nested within the prediction system, which are in turn nested within the communication platform. This hierarchical nesting allows complex intelligent prediction functionality to be integrated into existing platforms without requiring complete system redesign, thus improving automation while managing complexity through modular layering
Solution Approach 2:
The patent introduces AI models and prediction algorithms as intermediary components between users and the communication platform. These intermediaries process communication data, generate predictions, and provide analytics without replacing the core communication functionality, thereby adding intelligent capabilities while maintaining system manageability through clear separation of concerns
2Loss of time
If communication sessions are analyzed in real-time, then actionable insights are provided promptly, but processing time and computational resources increase
Solution Approach 1:
The patent performs preliminary processing of communication sessions by pre-training language models on large datasets and pre-processing communication transcripts during or after sessions. This preliminary action prepares data structures and models in advance, enabling faster real-time prediction and analytics generation without excessive computational overhead during actual analysis
Solution Approach 2:
The patent applies partial analysis by focusing computational resources on identifying and analyzing specific next step sentences rather than processing entire communication sessions uniformly. This selective approach reduces overall computational resource consumption while still delivering timely actionable insights for critical portions of communications
Data Source
AI summary
Methods and systems provide for extracting next step sentences from a communication session. In one embodiment, the system defines a set of annotation guidelines for labeling training data; receives a set of labeled training data including sentences from a transcript of a communication session, a subset of the sentences being associated with a positive label; organizes the labeled training data and trains a model with the labeled training data, the training including, for each of the sentences, inputting the sentence into a language model and a classification head to output a number of class probabilities, and inputting a classification token representing the sentence into a classification head; using a number of classifiers from the trained model to generate ensemble class scores; and using the ensemble class scores to predict one or more next step sentences from the sentences in the transcript.