# Amharic News Classification Using LSTM, Bidirectional-LSTM, and GRU
### I have collected 24,000 news from different news sources in Ethiopian media.
### Then I categorized the dataset into 6 classes. namely:
### 1. የሃገር ዉስጥ ዜና (Local News)
#### 2. ፖለቲካ (Politics News)
#### 3. ቢዝነስ (Business News)
#### 4. አለምአቀፋዊ ዜና (International News)
#### 5. መዝናኛ (Entertainment News)
#### 6. ስፖርት (Sport News)
## Data Preprocessing:
I have applied different data-preprocessing methods to make the dataset clean. such as
### 1. Removing any missing value from the whole dataset
### 2. Normalization
To normalize character level mismatch such as "ጸሀይ" and "ፀሐይ" and others.
#### 3. Removing Special Characters
Removing any special character from the article column.
### 4. Removing Stopwords
Removing the stopwords from the article column using the set Amharic stopwords
# Models Used:
### 1. LSTM (Long Short Term Memory)
### 2. Bidirectional LSTM
### 3. Gated Recurrent Unit
### 4. Word2vec
## N.B: if you need a dataset, contact me using LinkedIn
linkedin.com