Domain Specific Data Distillation and Multi-modal Embedding Generation

TL;DR

This paper presents a novel hybrid modeling approach that enhances domain-specific embeddings by filtering noise from unstructured data using structured data, leading to improved attribute prediction in the cloud computing domain.

Contribution

It introduces a hybrid collaborative filtering framework that fine-tunes entity representations with relevant item prediction, outperforming traditional autoencoder methods.

Findings

28% increase in precision for attribute prediction

11% increase in recall for attribute prediction

Effective noise filtering from unstructured data

Abstract

The challenge of creating domain-centric embeddings arises from the abundance of unstructured data and the scarcity of domain-specific structured data. Conventional embedding techniques often rely on either modality, limiting their applicability and efficacy. This paper introduces a novel modeling approach that leverages structured data to filter noise from unstructured data, resulting in embeddings with high precision and recall for domain-specific attribute prediction. The proposed model operates within a Hybrid Collaborative Filtering (HCF) framework, where generic entity representations are fine-tuned through relevant item prediction tasks. Our experiments, focusing on the cloud computing domain, demonstrate that HCF-based embeddings outperform AutoEncoder-based embeddings (using purely unstructured data), achieving a 28% lift in precision and an 11% lift in recall for domain-specific attribute prediction.