updated readme

dhakalmahima188 · Apr 13, 2024 · 501acb0 · 501acb0
1 parent 6c5ea70
commit 501acb0
Show file tree

Hide file tree

Showing 19 changed files with 124 additions and 56 deletions.
diff --git a/Results/AudioRecord.png b/Results/AudioRecord.png
diff --git a/Results/afb.png b/Results/afb.png
diff --git a/Results/afbi.png b/Results/afbi.png
diff --git a/Results/complainportal.png b/Results/complainportal.png
diff --git a/Results/datset.png b/Results/datset.png
diff --git a/Results/ewc.png b/Results/ewc.png
diff --git a/Results/grtelecom.png b/Results/grtelecom.png
diff --git a/Results/homepage.png b/Results/homepage.png
diff --git a/Results/lstm.png b/Results/lstm.png
diff --git a/Results/lstm2.png b/Results/lstm2.png
diff --git a/Results/mvc.png b/Results/mvc.png
diff --git a/Results/resultportal.png b/Results/resultportal.png
diff --git a/Results/sytem_design_final.png b/Results/sytem_design_final.png
diff --git a/Results/tvb.png b/Results/tvb.png
diff --git a/Results/tvbi.png b/Results/tvbi.png
diff --git a/Results/tvm.png b/Results/tvm.png
diff --git a/Results/wordcloud.png b/Results/wordcloud.png
diff --git a/image.png b/image.png
diff --git a/readMe.md b/readMe.md
@@ -1,56 +1,124 @@
-### Initial Setup
-
-#### Environment setup
-```python
-python3 -m venv env
-source bin/activate
-django-admin createproject major
-python3 manage.py startapp backend
-```
-#### Templates and Static
-```python
-'DIRS': [BASE_DIR,"templates"]
-STATICFILES_DIRS = [ os.path.join(BASE_DIR, "static")]
-```
-
-```html
-{% load static %}
- <link type="text/css" rel="stylesheet" href="{% static 'css/homepage.css' %}">
-```
-
-#### Database Setup
-
-
-```shell
-sudo apt install postgresql postgresql-contrib
-sudo systemctl start postgresql.service
-sudo -u postgres psql
-sudo apt install pgadmin4
-
-```
-
-
-```shell
-sudo -i -u postgres
-psql
-createdb mydb
-\l
-psql -d mydb
-\password postgres
-
-
-```
-
-```json
-DATABASES = {
-    'default': {
-        'ENGINE': 'django.db.backends.postgresql',
-        'NAME': 'mydb', 
-        'USER': 'postgres', 
-        'PASSWORD': 'postgres',
-        'HOST': 'localhost', 
-        'PORT': 5432,
-    }
-}
-
-```
+
+
+# Grievance Recognition System Based on Text/Speech
+
+## Introduction
+This project develops an innovative Grievance Recognition System that utilizes advanced NLP and machine learning techniques to automate the processing of grievances via text and speech inputs. Aimed at improving efficiency in handling user grievances, the system is a part of our Major Project submission, Department of Electronics and Computer Engineering, Pulchowk Campus, Tribhuvan University.
+
+## Project Team
+- Ashwini Mandal
+- Mahima Dhakal
+- Nadika Poudel
+- Aashish Kumar Sah
+
+## Supervisor
+- Dr. Basanta Joshi, Assistant Professor, Department of Electronics and Computer Engineering, Pulchowk Campus
+
+## System Overview
+The system is structured into three core components:
+1. **Automatic Speech Recognition (ASR)**: Utilizes the state-of-the-art wav2vec 2.0 model, fine-tuned for the Nepali language, to convert speech into text.
+2. **Text Classification**: Employs both traditional machine learning models and deep learning approaches including LSTM, BiLSTM, and transformer-based models like NepaliBERT and MuRIL for efficient classification of the grievances.
+3. **Web Interface**: A user-friendly web interface that facilitates easy submission and tracking of grievances.
+
+## Technologies Used
+- Python for backend processing.
+- PyTorch and Hugging Face Transformers for machine learning models.
+- HTML/CSS/JS and Bootstrap for frontend development.
+- MySQL for database management.
+
+## Setup and Installation
+Detailed instructions on how to set up and run the project locally.
+
+### Requirements
+  `requirements.txt `
+
+### Installation Steps
+1. Clone the repository 
+2. Install Python dependencies
+3. Setup Database
+4. Run the backend server : port
+5. Access the web interface at `http://localhost:port`.
+
+# Technical Description:
+
+## Project Overview
+The Grievance Recognition System is designed to automate the handling of grievances received in text or speech formats using advanced machine learning and natural language processing techniques. This system is especially tuned to understand and process inputs in the Nepali language, accommodating the linguistic nuances essential for effective grievance handling.
+
+## System Architecture
+
+### Components
+The system comprises three major components:
+1. **Automatic Speech Recognition (ASR)**
+2. **Text Classification**
+3. **Web Interface**
+
+### Automatic Speech Recognition (ASR)
+- **Model**: wav2vec 2.0, fine-tuned for Nepali language.
+- **Purpose**: Converts speech input into text. It allows users to submit their grievances orally, making the system accessible to a broader user base, including those unable to write.
+- **Process**:
+  - The speech audio is captured and processed using a feature extractor.
+  - The processed features are then fed into the wav2vec 2.0 model to generate the corresponding text output.
+
+### Text Classification
+- **Models Used**:
+  - Traditional ML models: SVM, Decision Trees.
+  - Deep Learning models: LSTM, BiLSTM.
+  - Transformer models: NepaliBERT, MuRIL.
+- **Purpose**: Classifies the text into categories such as complaints, queries, and suggestions, which can then be routed to the appropriate departments.
+- **Process**:
+  - Text input is tokenized and vectorized.
+  - The vectorized text is processed through the model to classify into predefined categories.
+
+### Web Interface
+- **Technologies**: Django
+- **Features**:
+  - Users can submit grievances through a text form or voice recording.
+  - Dashboard for tracking the status of submitted grievances.
+  - Admin panel for response management and analytics.
+  ![model](./Results/homepage.png)
+  ![model](./Results/complainportal.png)
+   ![model](./Results/AudioRecord.png)
+    ![model](./Results/resultportal.png)
+
+## Data Flow
+1. **Input Acquisition**: Users submit their grievances via the web interface using either text or speech.
+2. **Preprocessing**:
+   - **Speech to Text**: Audio inputs are converted to text using the ASR component.
+   - **Text Normalization**: Text inputs are cleaned and normalized for processing.
+3. **Classification**: The preprocessed text is then classified into categories using the text classification models.
+4. **Response Generation**: Based on the classification, automated responses or routing decisions are made.
+5. **Output Delivery**: Responses or status updates are communicated back to the user through the web interface.
+
+## Technologies
+- **Backend**: Python, Django, PyTorch, Hugging Face Transformers.
+  ![model](./Results/mvc.png)
+- **Frontend**: HTML-CSS, Bootstrap.
+- **Database**: MySQL.
+- **Machine Learning/Deep Learning**: scikit-learn, TensorFlow, PyTorch.
+- **Speech Processing**: Librosa, Hugging Face Transformers.
+
+## System Design
+  ![model](./Results/datset.png)
+   ![model](./Results/wordcloud.png)
+
+## Data
+ ![model](./Results/resultportal.png)
+## Model Results
+LSTM:
+![model](./Results/lstm.png)
+
+![model](./Results/lstm2.png)
+Nepbert :
+![model](./Results/afb.png)
+Bi-LSTM:
+![model](./Results/afbi.png)
+![model](./Results/tvbi.png)
+Muril:
+![model](./Results/tvm.png)
+
+## Continual learning
+![model](./Results/ewc.png)
+
+## Applications 
+Telecommunication :
+ ![model](./Results/grtelecom.png)