It’s a very common practice these days to search for information on the internet. Everything and anything you need is only one click away. But the results at first may not be exactly what you expected, and then we have to read those small paragraphs in each link just to understand what the page is really about. These paragraphs represent summaries of the main article. As the internet is loaded with millions of information every day from webpages, news, blogs, researches, etc. manually making a summary can be very difficult.
Google, Yahoo, and other search engines use tools for automatic text summarization in order to summarize all lengthy texts. A summarizer is primarily a system which takes sentences from a document, identifies the most relevant, and rearranges these in a structured and readable form, only much shorter. In natural language processing, automatic text summarization helps systems to analyse and fathom human language.
Two main techniques to automatic text summarization include:
- Extractive Method
2. Abstractive Method
The text summarization scope is determined based on input type, purpose, domain- or query-based, and output type.
The extractive method chooses phrases and sentences from an original document in order to create summaries. It ranks them in order of relevance, choosing the most relevant to the source document.
The abstractive method produces completely new sentences and phrases which reflect the essence of the original document. It offers more realistic results and is more challenging, but is used by humans. It works by choosing and reducing the content from the original, but it may produce words which are not in the original source document.
Abstractive method is thought to provide a general solution to the problem of abstraction, while extractive method is more effective and widely used because of its availability and easy approach.