Agent Task Scoring for New Task Types Without Historical Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing online systems face challenges in predicting which servicing agents are suited to perform new types of tasks without historical data, as traditional machine-learning models require training data that is not available for newly defined tasks.
Innovation Solution
A machine-learning model is trained to identify servicing agents' aptitude for new tasks using a database of tuples containing agent scores and task types, and applies this model to generate task scores for eligible agents, prompting them through user interfaces to accept new tasks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional machine-learning models are used to predict servicing agent performance, then prediction accuracy can be achieved for existing tasks, but the models cannot be trained for new task types due to lack of historical data
Solution Approach 1:
The system performs preliminary actions by collecting and storing diverse agent characteristics and performance data during normal operations. When a new task type is introduced, this pre-collected data serves as the foundation for immediate prediction without requiring historical data for the specific new task, thus resolving the cold start problem
Solution Approach 2:
The machine learning model is designed with universal applicability to handle multiple task types. By training the model on diverse task data and using task-agnostic agent characteristics, the same model architecture can predict performance across both existing and new task types, making the system versatile without requiring task-specific model development
2Adaptability or versatility
If new types of tasks are introduced to increase agent work flexibility and earnings potential, then agent satisfaction improves, but the system cannot effectively match agents to new tasks without historical performance data
Solution Approach 1:
The system introduces an intermediary matching layer that translates between agent characteristics and new task requirements. This intermediary uses the machine learning model to bridge the information gap, enabling effective task assignment even when direct historical performance data for the new task type is unavailable
Solution Approach 2:
The system performs preliminary data collection and agent profiling during normal operations with existing tasks. This pre-established knowledge base of agent capabilities serves as a foundation for immediately assigning new task types, preventing information loss when transitioning to novel tasks
3Measurement precision
If traditional rating approaches are applied across various task types, then existing tasks can be evaluated, but new task types cannot be properly assessed due to fundamental differences from existing tasks
Solution Approach 1:
The system dynamically adjusts evaluation parameters based on task characteristics. Rather than applying a fixed rating approach, the machine learning model adapts the evaluation criteria to match the specific requirements of each task type, whether existing or new, maintaining measurement precision across diverse tasks through parameter transformation
Data Source
AI summary
An online system uses a machine-learning model to identify servicing agents suited to perform tasks of new types. The online system maintains a list of tuples for servicing agents, each tuple including a score for a servicing agent and an identifier of a task type, the score indicating a level of aptitude of the servicing agent to perform a task of the task type. Upon obtaining a description for a task of a new type, the online system applies the machine-learning model to the list of tuples and the description for the task to generate a task score for each servicing agent that is indicative of a level of aptitude of each servicing agent for performing the task of the new type. The online system selects, using the task score for each servicing agent, servicing agents to whom the online system offers the task of the new type.