From 18c4ea42f487a7e1f675e456ec57fd1df4e62ebb Mon Sep 17 00:00:00 2001
From: Muhammed Enes Ekinci <211805030@stu.adu.edu.tr>
Date: Tue, 27 Aug 2024 21:31:31 +0300
Subject: [PATCH] readme

---
 README.md | 39 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 39 insertions(+)
 create mode 100644 README.md

diff --git a/README.md b/README.md
new file mode 100644
index 0000000..c19deca
--- /dev/null
+++ b/README.md
@@ -0,0 +1,39 @@
+# News Generation
+
+## Introduction
+
+This project is a part of the Teknofest 2024 Türkçe Doğal Dil İşleme competition. The aim of the project is to
+generate news title and content from a given image.
+
+## Dataset
+
+The dataset is collected from the ![Sabah](https://www.sabah.com.tr/timeline/) news website.
+The dataset consist of news titles, news content and images. The dataset is in Turkish Language.
+
+## Data-Preprocessing
+
+- Sample Data:
+  ![image](https://isbh.tmgrup.com.tr/sbh/2022/03/31/balikesirde-tarihi-binada-baslayan-ve-restorana-sicrayan-yangin-sonduruldu-1648686422986.jpeg)
+
+  ***
+
+  ```python
+    title = "Balıkesir’de tarihi bina yangında küle döndü"
+    word_index = {'Balıkesir’de': 9, 'tarihi': 5, 'bina': 3, 'yangında': 7, 'küle': 5, 'döndü': 6 }
+    tokens: [start_token, 9, 5, 3, 7, 5, 6, end_token]
+  ```
+
+  | Input                                       | Output    |
+  | ------------------------------------------- | --------- |
+  | Image + start_token                         | 9         |
+  | Image + start_token + 9                     | 5         |
+  | Image + start_token + 9 + 5                 | 3         |
+  | Image + start_token + 9 + 5 + 3             | 7         |
+  | Image + start_token + 9 + 5 + 3 + 7         | 5         |
+  | Image + start_token + 9 + 5 + 3 + 7 + 5     | 6         |
+  | Image + start_token + 9 + 5 + 3 + 7 + 5 + 6 | end_token |
+
+## Model
+
+The model is a combination of CNN and LSTM, where the image is fed to the Encoder(CNN) and the output of the CNN is
+fed to the Decoder(LSTM) along with the input text.