# bintanng99/fake-job-posting-detection-bert architecture

> Architecture diagram of the GitHub repository bintanng99/fake-job-posting-detection-bert, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-09-30). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/bintanng99/fake-job-posting-detection-bert
- Repository: https://github.com/bintanng99/fake-job-posting-detection-bert

## Overview

This repository is an exploratory notebook for binary fraudulent-job-posting classification, not a completed screening system. Its documented workflow cleans and combines posting fields, splits the data, undersamples only the training set, tokenizes with BERT, and defines three frozen-BERT classifier heads. The saved run starts training the LSTM variant but has no recorded test results or completed architecture comparison. The notebook itself was not excerpted, so relationships below rely on the README’s workflow description.

Main entry points: [`fake_job_posting_detection_bert.ipynb`](https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb) (Model training)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_preparation["Data preparation"]
  node_dataset[("Local dataset")]
  node_cleaning["Text cleaning"]
  node_combination["Text combination"]
  node_split["Data splitting"]
  node_undersampling["Training undersampling"]
end

subgraph group_modeling["BERT modeling"]
  node_tokenization["BERT tokenization"]
  node_bert["Frozen BERT"]
  node_lstm["BiLSTM head"]
  node_cnn["CNN head"]
  node_combined["LSTM-CNN head"]
  node_training["Model training"]
end

subgraph group_assessment["Experiment assessment"]
  node_evaluation["Metric evaluation"]
  node_artifact["Model artifact"]
end

node_researcher(("Researcher"))

node_researcher -->|"provides"| node_dataset
node_dataset -->|"loads and cleans"| node_cleaning
node_cleaning -->|"prepares fields"| node_combination
node_combination -->|"feeds"| node_split
node_split -->|"selects training"| node_undersampling
node_undersampling -->|"feeds"| node_tokenization
node_tokenization -->|"encodes tokens"| node_bert
node_bert -->|"feeds head"| node_lstm
node_bert -->|"feeds head"| node_cnn
node_bert -->|"feeds head"| node_combined
node_lstm -->|"trains first"| node_training
node_cnn -->|"candidate for"| node_training
node_combined -->|"candidate for"| node_training
node_training -.->|"intended evaluation"| node_evaluation
node_training -.->|"intended save"| node_artifact

click node_cleaning "https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb"
click node_combination "https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb"
click node_split "https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb"
click node_undersampling "https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb"
click node_tokenization "https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb"
click node_bert "https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb"
click node_lstm "https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb"
click node_cnn "https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb"
click node_combined "https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb"
click node_training "https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb"
click node_evaluation "https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb"
click node_artifact "https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_dataset,node_cleaning,node_combination,node_split,node_undersampling toneBlue
class node_tokenization,node_bert,node_lstm,node_cnn,node_combined,node_training toneAmber
class node_evaluation,node_artifact toneMint
class node_researcher toneIndigo
```

## Components

### Data preparation

- **Local dataset**
- [Text cleaning](https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb): `fake_job_posting_detection_bert.ipynb`
- [Text combination](https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb): `fake_job_posting_detection_bert.ipynb`
- [Data splitting](https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb): `fake_job_posting_detection_bert.ipynb`
- [Training undersampling](https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb): `fake_job_posting_detection_bert.ipynb`

### BERT modeling

- [BERT tokenization](https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb): `fake_job_posting_detection_bert.ipynb`
- [Frozen BERT](https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb): `fake_job_posting_detection_bert.ipynb`
- [BiLSTM head](https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb): `fake_job_posting_detection_bert.ipynb`
- [CNN head](https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb): `fake_job_posting_detection_bert.ipynb`
- [LSTM-CNN head](https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb): `fake_job_posting_detection_bert.ipynb`
- [Model training](https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb): `fake_job_posting_detection_bert.ipynb`

### Experiment assessment

- [Metric evaluation](https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb): `fake_job_posting_detection_bert.ipynb`
- [Model artifact](https://github.com/bintanng99/fake-job-posting-detection-bert/blob/main/fake_job_posting_detection_bert.ipynb): `fake_job_posting_detection_bert.ipynb`

### Other components

- **Researcher**

## Connections

- Researcher → Local dataset: provides (evidence: `README.md`)
- Local dataset → Text cleaning: loads and cleans (evidence: `README.md`)
- Text cleaning → Text combination: prepares fields (evidence: `README.md`)
- Text combination → Data splitting: feeds (evidence: `README.md`)
- Data splitting → Training undersampling: selects training (evidence: `README.md`)
- Training undersampling → BERT tokenization: feeds (evidence: `README.md`)
- BERT tokenization → Frozen BERT: encodes tokens (evidence: `README.md`)
- Frozen BERT → BiLSTM head: feeds head (evidence: `README.md`)
- Frozen BERT → CNN head: feeds head (evidence: `README.md`)
- Frozen BERT → LSTM-CNN head: feeds head (evidence: `README.md`)
- BiLSTM head → Model training: trains first (evidence: `README.md`)
- CNN head → Model training: candidate for (evidence: `README.md`)
- LSTM-CNN head → Model training: candidate for (evidence: `README.md`)
- Model training → Metric evaluation: intended evaluation
- Model training → Model artifact: intended save
