Automatic summarization is the use of a computer program to create a summary of a text or texts. This can be useful in a variety of settings, including document searching, education and research. Programs can approach this challenge in a number of ways. Computer scientists and other researchers who have an interest in natural language have studied ways to develop automatic summarization software to improve the quality of services available to users of such software.
One approach to automatic summarization involves a quick scan of the document to identify the most important information. The program learns how to find important content by looking at wording, context and presentation. It might look for material such as the abstract on a lab report or the first line definition in an encyclopedia article. Next, it can lift out key sentences and use them to create a summary by presenting these copies, as seen with many search engines.
A more sophisticated approach is the actual creation of an abstract. In this case, the computer program reviews the text, synthesizes the information and presents a condensed version to the user. This type of automatic summarization requires more advanced programming. The computer does not just need to find the most important information, it needs to present it in new wording for the benefit of the user.
As a search tool, automatic summarization can be extremely valuable. Many Internet users rely on the quick extracts provided on a list of search results, for instance, to determine which articles are relevant to their needs. Scanning these excerpts can help the user decide whether to click on the link. Abstracts can be useful for people such as researchers who want a quick overview of discussion on a particular topic. If a particular abstract is especially interesting, they can click through to read the piece in its entirety.
Adaptable software might be able to learn through automatic summarization. The reader can rate summaries in terms of how useful they are and whether they accurately convey the information in the source text. This allows the program to pick up on where it might have gone wrong. It can use this information to improve the quality and accuracy of results in the future. Developers who have an interest in automatic summarization might engage in activities such as experiments to pit humans and machines against each other to see which can come up with the most appropriate summaries of texts.
