Temporal Context-Aware Question Routing Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Community-based question answering systems face challenges in efficiently routing questions to the most relevant answerers due to multifaceted expertise and temporal dynamics, where existing approaches fail to accurately model users' expertise and recent answering behavior.
Innovation Solution
A temporal context-aware question routing model (TCQR) is introduced, which encodes answerers into representations based on semantic and temporal information, using multi-shift and multi-resolution extensions to model answering behaviors across different time granularities, and employs triplet loss functions to learn users' answering behavior.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing approaches learn users' expertise from their past question-answering activities, then the system can identify potential answerers, but it fails to accurately model multi-faceted expertise and temporal dynamics in answering behavior
Solution Approach 1:
The patent segments the temporal dynamics modeling into multiple time granularities (short-term, medium-term, long-term) using separate LSTM networks for each granularity level. This allows the system to capture answering behaviors at different temporal scales simultaneously, resolving the contradiction between measurement precision and adaptability by handling temporal dynamics through structured segmentation rather than a single monolithic model.
Solution Approach 2:
The patent adds a temporal dimension to the expertise modeling by incorporating time-aware LSTM networks that process historical answering behaviors across multiple time granularities. This transforms the static expertise representation into a dynamic, time-aware representation, enabling the system to accurately measure expertise while adapting to temporal dynamics in answering behaviors.
2Measurement precision
If the system considers multiple time granularities for temporal dynamics, then the modeling accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent divides the temporal modeling task into segmented LSTM networks operating at different time granularities (short-term, medium-term, long-term). Each segmented network focuses on a specific temporal scale, which improves modeling accuracy while managing complexity through modular architecture rather than requiring a single complex network to handle all temporal scales.
Solution Approach 2:
The patent merges the outputs from multiple segmented LSTM networks (operating at different time granularities) into a unified temporal context representation. This combining approach integrates information from multiple time scales to improve overall modeling accuracy while maintaining manageable complexity through the modular segmented architecture that processes different granularities independently before integration.
3Measurement precision
If the system encodes answerers based on semantic and temporal information, then the question routing accuracy improves, but the processing time increases
Solution Approach 1:
The patent performs preliminary encoding of answerers into temporal context-aware representations using LSTM networks that process historical answering behaviors. By pre-computing and storing these temporal embeddings, the system reduces processing time during actual question routing operations, as the complex temporal analysis has already been performed in advance rather than being computed in real-time for each routing decision.
Solution Approach 2:
The patent implements dynamic temporal context-aware encoding that adapts to different time granularities and answering behaviors. The LSTM-based encoding dynamically adjusts to capture relevant temporal patterns, improving routing accuracy while the dynamic nature of the encoding allows efficient processing by focusing computational resources on the most relevant temporal features rather than processing all possible temporal information uniformly.
Data Source
AI summary
A method for employing a temporal context-aware question routing model (TCQR) in multiple granularities of temporal dynamics in community-based question answering (CQA) systems is presented. The method includes encoding answerers into temporal context-aware representations based on semantic and temporal information of questions, measuring answerers expertise in one or more of the questions as a coherence between the temporal context-aware representations of the answerers and encodings of the questions, modeling the temporal dynamics of answering behaviors of the answerers in different levels of time granularities by using multi-shift and multi-resolution extensions, and outputting answers of select answerers to a visualization device.