Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

CTG-KrEW: Generating Synthetic Structured Contextually Correlated Content by Conditional Tabular GAN with K-Means Clustering and Efficient Word Embedding

Riya Samanta, Bidyut Saha, Soumya K. Ghosh, Sajal K. Das

TL;DR

CTG-KrEW tackles the challenge of generating realistic synthetic tabular data with contextually coherent word sequences, such as skillsets, by introducing a three-step preprocessing pipeline: unique word extraction, word2vec encoding, and K-Means clustering. This encoded representation, fed into a CTGAN core (with a specialized architecture and training regime), preserves semantic relationships among words while remaining computationally efficient. The framework demonstrates substantial improvements in efficiency and data quality on UpWork-related datasets, including lower memory and CPU requirements and closer alignment to source distributions as evidenced by refined entropy, KL-divergence, and associativity metrics. A publicly accessible KrEW web application further enables scalable generation of synthetic worker/task data, highlighting practical utility for privacy-preserving data synthesis across domains beyond skill-centric contexts.

Abstract

Conditional Tabular Generative Adversarial Networks (CTGAN) and their various derivatives are attractive for their ability to efficiently and flexibly create synthetic tabular data, showcasing strong performance and adaptability. However, there are certain critical limitations to such models. The first is their inability to preserve the semantic integrity of contextually correlated words or phrases. For instance, skillset in freelancer profiles is one such attribute where individual skills are semantically interconnected and indicative of specific domain interests or qualifications. The second challenge of traditional approaches is that, when applied to generate contextually correlated tabular content, besides generating semantically shallow content, they consume huge memory resources and CPU time during the training stage. To address these problems, we introduce a novel framework, CTGKrEW (Conditional Tabular GAN with KMeans Clustering and Word Embedding), which is adept at generating realistic synthetic tabular data where attributes are collections of semantically and contextually coherent words. CTGKrEW is trained and evaluated using a dataset from Upwork, a realworld freelancing platform. Comprehensive experiments were conducted to analyze the variability, contextual similarity, frequency distribution, and associativity of the generated data, along with testing the framework's system feasibility. CTGKrEW also takes around 99\% less CPU time and 33\% less memory footprints than the conventional approach. Furthermore, we developed KrEW, a web application to facilitate the generation of realistic data containing skill-related information. This application, available at https://riyasamanta.github.io/krew.html, is freely accessible to both the general public and the research community.

CTG-KrEW: Generating Synthetic Structured Contextually Correlated Content by Conditional Tabular GAN with K-Means Clustering and Efficient Word Embedding

TL;DR

CTG-KrEW tackles the challenge of generating realistic synthetic tabular data with contextually coherent word sequences, such as skillsets, by introducing a three-step preprocessing pipeline: unique word extraction, word2vec encoding, and K-Means clustering. This encoded representation, fed into a CTGAN core (with a specialized architecture and training regime), preserves semantic relationships among words while remaining computationally efficient. The framework demonstrates substantial improvements in efficiency and data quality on UpWork-related datasets, including lower memory and CPU requirements and closer alignment to source distributions as evidenced by refined entropy, KL-divergence, and associativity metrics. A publicly accessible KrEW web application further enables scalable generation of synthetic worker/task data, highlighting practical utility for privacy-preserving data synthesis across domains beyond skill-centric contexts.

Abstract

Conditional Tabular Generative Adversarial Networks (CTGAN) and their various derivatives are attractive for their ability to efficiently and flexibly create synthetic tabular data, showcasing strong performance and adaptability. However, there are certain critical limitations to such models. The first is their inability to preserve the semantic integrity of contextually correlated words or phrases. For instance, skillset in freelancer profiles is one such attribute where individual skills are semantically interconnected and indicative of specific domain interests or qualifications. The second challenge of traditional approaches is that, when applied to generate contextually correlated tabular content, besides generating semantically shallow content, they consume huge memory resources and CPU time during the training stage. To address these problems, we introduce a novel framework, CTGKrEW (Conditional Tabular GAN with KMeans Clustering and Word Embedding), which is adept at generating realistic synthetic tabular data where attributes are collections of semantically and contextually coherent words. CTGKrEW is trained and evaluated using a dataset from Upwork, a realworld freelancing platform. Comprehensive experiments were conducted to analyze the variability, contextual similarity, frequency distribution, and associativity of the generated data, along with testing the framework's system feasibility. CTGKrEW also takes around 99\% less CPU time and 33\% less memory footprints than the conventional approach. Furthermore, we developed KrEW, a web application to facilitate the generation of realistic data containing skill-related information. This application, available at https://riyasamanta.github.io/krew.html, is freely accessible to both the general public and the research community.
Paper Structure (12 sections, 6 equations, 30 figures, 7 tables, 1 algorithm)

This paper contains 12 sections, 6 equations, 30 figures, 7 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Dataset with Sequentially Correlated Word Attributes
  4. Preliminaries
  5. Generic CTGAN Approach with Default Encoding
  6. Generic CTGAN Approach with Multi-hot Encoding
  7. Proposed CTG-KrEW Framework
  8. Core Architecture of CTGAN
  9. Evaluation
  10. KrEW Web Application
  11. Discussion and Use-Cases
  12. Conclusion

Figures (30)

  • Figure 1: Snapshots of task-data showing only 20 rows, outsourced from Kaggle crawlfeed
  • Figure 3: Skill distribution in the task-data of the source and synthetic datasets in 3D space with coordinates in the range $[-2,2]$, $[0,5]$, and $[-0.5,2]$ by using PCA.
  • Figure 5: Workflow of CTG-KrEW
  • Figure 6: task-data
  • Figure 7: worker-data
  • ...and 25 more figures