NLP-project-Mental-Graph-Of-Users-Using-Social-Media-Posts-

Mental Graph Of Users Using Social Media Posts - g2

Methodology

Extracting Twitter Tweets & Youtube Comments from videos.
Classifying positive and negative sentiments from extracted data.
Among the negative sentiments classifying them further into hate and criticism posts.
Plotting a bar graph of different negative sentiment from a user.

Steps Of Project

Data Extraction & Labelling
Data Preprocessing
Data Visualization
Word Embeddings
Models
Results
Model Deployment

Data Extraction & Labelling

Extraction:

Made use of snscrape library and Youtube API for extracting Twitter and YouTube data respectively.

Labelling:

Performed manual labelling for first classification into 2 classes i.e Positive , Negative.
Further performed manual labelling for second classification into 2 classes i.e. Hate and Criticism .

Data Preprocessing

Removed Urls from tweets
Removed hashtags and emoticons
Removed special characters
Removed Retweet markers
Removed Stopwords
Applied Lemmatization

Word Embeddings

Used word2vec word embeddings.
Embedding Dimension: 200 .
Vocab Size: 1600 .

Models Implemented

Two models applied one for sentiment analysis and other for hate and criticism classification

Bidirectional LSTM
Small Bert

Future Scope

Making use of huge data.
Adding more project specific cleaning steps.
Adding hindi language to the model.
Finding better word embedding algorithm.
Finding better classification model.
Developing more interactive user interface.
In short, to make a project complex enough to actually help the officials to track down targeted people.

itsvikaspandey/NLP-project-Mental-Graph-Of-Users-Using-Social-Media-Posts-