Extracting Tasks from Customize Portal using Natural Language Processing

Abstract

In software documentation, product knowledge and software requirement are very important to improve product quality. Within maintenance stage, reading of whole documentation of large corpus won’t be possible by developers. They need to receive software documentation i.e. (development, designing and testing etc.) in a short period of time. Important documents are able to record in software documentation. There live a space between information which developer wants and software documentation. To solve this problem, an approach for extracting relevant task that is based on heuristically matching the structure of the documentation under three phases of software documentation (i.e. documentation, development and testing) is described. Our main idea is that task is extracted automatically from the software documentation, freeing the developer easily get the required data from software documentation with customize portal using WordNet library and machine learning technique. And then the category of task can be generated easily from existing applications using natural language processing. Our approach use WordNet library to identify relevant tasks for calculating frequency of each word which allows developers in a piece of software to discover the word usage.