Text summarization is a process of shortening a text document in a way that it contains the main points of the original document.

• reduces reading time
• lorem ipsum
• lorem ipsum

• Extraction: When the sentences of the summary are selected from the original document.
• Abstraction: When the model creates new sentences and forms a summary. Humans summarize documents mostly based on abstractive method.

• Generic: Model tries to make a general summary of the document and treats all inputs the same.
• Query-Based: Model summarizes based on a natural language question.

• Single Document: Model creates the summary based on 1 paper or document.
• Multiple Document: Model generates the summary based on multiple papers about a specific subject.

The main idea in this approach is to find sentence embeddings by averaging vector representations of the words of sentence and assign the cosine similarity of those vectors to weights of a fully connected graph. Then perform a pagerank algorithm on this graph and chose the sentences with higher rank as the sentences of our extractive summary.