원문정보
초록
영어
In the past years, the scale of software is growing quickly as more and more organizations begin to deploy their business on Internet. As a result, requirement analysis becomes a challenging issue and conventional approaches might significantly increase the costs of software development. Therefore, automatic requirement analysis techniques have attract more and more attentions, which allows for modeling and analyzing requirements formally, rapidly and automatically, avoiding mistakes made by misunderstanding between engineers and users, and saving lots of time and manpower. In this paper, we propose an approach to acquiring requirements automatically, which adopts automated planning techniques and machine learning methods to convert software requirement into an incomplete planning domain. By this approach, we design an algorithm called Intelligent Planning based Requirement Analysis (IPRA), to learn action models with uncertain effects. A concrete experiment is conducted to investigate the proposed algorithm, and the results indicate that it can obtain a complete planning domain and convert it into software requirement specification.
목차
1. Introduction
2. Related Work
3. Problem Description
4. Framework of IPRA Algorithm
4.1. Encode Plan Traces
4.2. Generate Candidate Formulas of Actions
4.3. Learn Weights of Candidate Formulas
4.4. Obtain Action Model
5. Experiments and Evaluation
5.1. Settings
5.2. Accuracy and Observed Intermediate States
5.3. Plan Traces in Action-Model Learning
6. Conclusion
References