There are many cases in which entrepreneurs need to have practical insights from collections of journal articles, company documentation and other relevant text documents to co-create solutions to entrepreneurial problems focusing on scaling companies early, rapidly and securely. Therefore they need text mining machine learning techniques to generate competitive insights from online textual data. Creating the summary of documents physically is a tedious job. Therefore, text summarization, a technique in natural language processing for constructing a small version of a large document, is usually used, and there are two main techniques:
Extractive summarization Picking vital sentences from the original document
Abstractive Summarization Consuming semantic information from the original document and producing the summary with completely new sentences
The abstractive summarization approach can be mainly categorized into two methods. Structured based approaches It uses deep learning techniques to choose the critical parts from the source documents. Semantic-based approaches. Multimodal semantic model: In this technique, a semantic unit, which extracts the subject matter and correlation among the topic, is created to represent the topic (pictures and text data) of one or more documents. Semantic graph-based method: This technique creates a summary by building the semantic graph called Rich Semantic Graph (RSG) on the source document, condensing the semantic graph, and making the complete abstractive summary from the reduced semantic graph