You can automatically extract highly informative chunks of text to create a meaningful and coherent summary. To accommodate for various use cases, you can control the compression ratio of the input text to extract summaries of various size. For example, to extract only a few most important sentences from large bodies of text, such as business emails or review articles, you can specify the compression ratio as 10% of the original text.
The following figure shows the summarization settings in a Text Analyzer rule.
Each summarization result consists of a number of sentences that the internal algorithm selected as the ones that carry the most meaning and information in relation to the length of the input text and the compression ratio setting. Each of these sentences is represented in your application as the pyExtractedSummary property with a subscript value.
The beginning and end of each sentence are marked as pyBegin and pyEnd integer properties that represent the place in the sequence of characters within the input string. The following figure illustrates example input text, items in the pyExtractedSummary list, and the corresponding summary result:
The summarization feature can help you automatically extract the text context by combining summarization with topic and intent detection, entity extraction, and sentiment analysis. For example, by extracting the context from the text, customer service representatives or business executives can make business decisions without the need to read the full text of a lengthy email, article, comment, and so on.
For more information, see: