Building Recommendation Engine for Cold-Start Model Switching

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Building recommendation systems face challenges when little to no user data is available, leading to a period where recommendations are unavailable, as they need to wait for sufficient data to train a single recommendation model, which can take months or years, and using a single technique may not yield the best results.

Innovation Solution

A building system that transitions between different recommender models based on feedback data, starting with an evolving matrix method (EMM) and potentially switching to collaborative filtering (CF), supervised learning (SL), and reinforcement learning (RL) models as performance is compared to an evolving reference model, allowing for immediate recommendation generation even with limited data.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If a single recommendation model is used, then the system is simple to implement, but it cannot generate recommendations when little data is available and may not yield the best results

Engineering Contradiction:

Improveease of implementationVSAvoidrecommendation generation capability

Core Design Contradiction:

Ease of manufactureVSProductivity

Solution Approach 1:

The system segments the recommendation model into multiple specialized models (collaborative filtering, content-based, hybrid) that can be independently selected and activated based on data availability and performance metrics, allowing each model to excel in specific scenarios

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system dynamically transitions between different recommendation models based on real-time performance evaluation and data availability, switching from single-model to multi-model approaches as the building accumulates more operational data

Inventive Principle:

Principle #15Dynamics

2Device complexity

If a single recommendation model is used, then the system complexity is low, but the recommendation quality may not be optimal

Engineering Contradiction:

Improvesystem complexityVSAvoidrecommendation quality

Core Design Contradiction:

Device complexityVSReliability

Solution Approach 1:

The system changes the operational parameters of the recommendation engine by adjusting which models are active based on data maturity stages, transitioning from simplified single-model operation to sophisticated multi-model ensembles as more data becomes available

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If the system waits for sufficient data to train a recommendation model, then the model accuracy improves, but the system cannot provide recommendations during the data collection period

Engineering Contradiction:

Improvemodel accuracyVSAvoiddata collection waiting time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary actions by implementing multiple pre-configured recommendation models that can operate immediately upon deployment, eliminating the cold-start problem and allowing recommendation generation to begin before extensive data collection occurs

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system introduces intermediary models (such as content-based filtering using building metadata and rules-based systems) that can generate recommendations without requiring extensive training data, bridging the gap until sufficient data accumulates for more sophisticated models

Inventive Principle:

Principle #24Intermediary (Mediator)

4Reliability

If multiple recommendation models are used, then the recommendation quality improves, but the system complexity increases

Engineering Contradiction:

Improverecommendation qualityVSAvoidsystem complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system achieves universality by designing a unified recommendation engine architecture that can accommodate multiple different model types and switching strategies, allowing the same system to handle diverse recommendation scenarios without requiring separate systems

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS20210383276A1Building system with a recommendation engine that operates without starting data

Publication Date: 2021.12.09 TYCO FIRE & SECURITY GMBH

AI summary

A building system including one or more memory devices configured to store instructions thereon, that, when executed by one or more processors, cause the one or more processors to generate building recommendations based on recommendation requests with a model of a first model type when less than a predefined amount of model training data is available and receive feedback data on the recommendation requests generated by the model of the first model type. The instructions cause the one or more processors to transition from generating the building recommendations by the model of the first model type to a second model of a second model type by comparing performance of the first model type to the second model type based on the feedback data.