{"id":18249556,"url":"https://github.com/ivanrj7j/ner-bilstm","last_synced_at":"2026-05-02T03:08:00.779Z","repository":{"id":246371982,"uuid":"820936020","full_name":"ivanrj7j/NER-BiLSTM","owner":"ivanrj7j","description":"This project showcases a custom Named Entity Recognition (NER) model using Bidirectional Long Short-Term Memory (BiLSTM) layers for precise entity predictions. The model is trained using a specific dataset and is integrated with a simple user interface built with HTML, CSS, and Flask.","archived":false,"fork":false,"pushed_at":"2024-06-27T14:06:57.000Z","size":74069,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":1,"default_branch":"main","last_synced_at":"2025-04-08T20:23:55.541Z","etag":null,"topics":["ai","lstm","ner","nlp","tensorflow"],"latest_commit_sha":null,"homepage":"","language":"Jupyter Notebook","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/ivanrj7j.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2024-06-27T13:25:47.000Z","updated_at":"2024-12-15T06:47:50.000Z","dependencies_parsed_at":null,"dependency_job_id":"0822e317-8c6c-4be9-a874-01665fd8cb35","html_url":"https://github.com/ivanrj7j/NER-BiLSTM","commit_stats":null,"previous_names":["ivanrj7j/ner-bilstm"],"tags_count":1,"template":false,"template_full_name":null,"purl":"pkg:github/ivanrj7j/NER-BiLSTM","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ivanrj7j%2FNER-BiLSTM","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ivanrj7j%2FNER-BiLSTM/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ivanrj7j%2FNER-BiLSTM/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ivanrj7j%2FNER-BiLSTM/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/ivanrj7j","download_url":"https://codeload.github.com/ivanrj7j/NER-BiLSTM/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ivanrj7j%2FNER-BiLSTM/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":32521123,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-02T01:12:54.858Z","status":"online","status_checked_at":"2026-05-02T02:00:05.923Z","response_time":132,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ai","lstm","ner","nlp","tensorflow"],"created_at":"2024-11-05T09:40:29.930Z","updated_at":"2026-05-02T03:08:00.758Z","avatar_url":"https://github.com/ivanrj7j.png","language":"Jupyter Notebook","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Custom Trained NER using Bidirectional LSTM\n\n## Introduction\nThis project showcases a custom Named Entity Recognition (NER) model using Bidirectional Long Short-Term Memory (BiLSTM) layers for precise entity predictions. The model is trained using a specific dataset and is integrated with a simple user interface built with HTML, CSS, and Flask.\n\n## Features\n- **Custom NER Model**: Utilizes Bidirectional LSTM layers for accurate predictions.\n- **User Interface**: Simple dark-themed UI for inputting text and viewing NER outputs.\n- **Pre-trained Model**: The project includes the pre-trained NER model, ready for use.\n\n## Technologies Used\n- TensorFlow\n- Flask\n- HTML \u0026 CSS\n\n## Dataset\nThe model is trained using the NER dataset provided by [Raj Nath Patel on Kaggle](https://www.kaggle.com/datasets/rajnathpatel/ner-data).\n\n## Project Motivation\nThe primary motivation for this project was to gain hands-on experience with machine learning and NER models, and to showcase these skills in a practical application.\n\n## Data Preprocessing\nThe provided dataset was already well-preprocessed, so the primary task was to create a custom data pipeline for feeding the data into the model.\n\n## Model Architecture\nThe NER model is built using a series of Bidirectional LSTM layers with dropout for regularization. Below is a table summarizing the model architecture:\n\n| Layer          | Output Shape       | Additional Parameters                               |\n|----------------|--------------------|-----------------------------------------------------|\n| Input          | (None, maxLen)     | `maxLen` = 35                                       |\n| Embedding      | (None, maxLen, 200)| `input_dim`=totalTokens, `output_dim`=200           |\n| Bidirectional LSTM (LSTM1) | (None, maxLen, 256)| `units`=128, `return_sequences`=True, `recurrent_dropout`=0.1 |\n| Bidirectional LSTM (LSTM2) | (None, maxLen, 256)| `units`=128, `return_sequences`=True, `recurrent_dropout`=0.1 |\n| Bidirectional LSTM (LSTM3) | (None, maxLen, 256)| `units`=128, `return_sequences`=True, `recurrent_dropout`=0.1 |\n| Dropout        | (None, maxLen, 256)| `rate`=0.5                                          |\n| TimeDistributed Dense (Classifier) | (None, maxLen, numClasses)| `units`=numClasses, `activation`=\"softmax\" |\n\n## Training Process\nThe model was trained for 10 epochs, which took approximately 15 minutes on the available hardware.\n\n## Evaluation Metrics\nThe model achieved a validation accuracy of 97.47%.\n\n## User Interface\nThe UI features a text input area for entering sentences and an output area for displaying the NER results. The labels are color-coded for easy identification of different entity types.\n\n![UI Screenshot](ui.png)\n\n## Prerequisites\n- TensorFlow\n\n## Installation and Setup\n1. Clone the repository:\n    ```sh\n    git clone https://github.com/ivanrj7j/NER-BiLSTM\n    ```\n2. Install the required packages:\n    ```sh\n    pip install -r requirements.txt\n    ```\n3. Navigate to the interface folder:\n    ```sh\n    cd interface\n    ```\n4. Run the application:\n    ```sh\n    python main.py\n    ```\n\n## Usage\n- Open a browser and go to `http://127.0.0.1:5000/`.\n- Enter text in the input area and click the \"Submit\" button.\n- View the NER results in the output area.\n\n## Known Issues\n- The model can only process up to 35 tokens at once, which may result in loss of context during prediction.\n\n## Acknowledgments\n- [Raj Nath Patel](https://www.kaggle.com/rajnathpatel) for the dataset.\n\n## License\nThis project is licensed under the MIT License.\n\n## Contact\nFor any questions or suggestions, feel free to contact me at:\n- **Email**: ivanrj7j@gmail.com\n- **GitHub**: [ivanrj7j](https://github.com/ivanrj7j)\n\n---\n\nFeel free to make any adjustments or let me know if there's anything more you'd like to add.","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fivanrj7j%2Fner-bilstm","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fivanrj7j%2Fner-bilstm","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fivanrj7j%2Fner-bilstm/lists"}