RUCMBot：Towards Efficient Use Case Modeling with Automated Domain Classification and Term Recommendation

• RUCMBot：Towards Efficient Use Case Modeling with Automated Domain Classification and Term Recommendation
  • 1. Contributions
  • 2. Overview
  • 3. Features of RUCMBot Tool
    • 1. Domain Classifier
    • 2. Domain Terminology Recommendation
    • 3. Domain Terminology Dictionary Auto-Construction

1. Contributions

1. Designing a model to classify domains automatically at run time when RUCM is used;
2. Extending OpenNLP Chunker for extracting domain terms to alleviate the problem of the low recall rate in extracting verb phrase (VP) chunks;
3. Developing a tool, named RUCMBot, which integrates with RUCM to enable efficient use case modeling.

2. Overview

An overview of RUCMBot is shown here. It contains two main components: recommending domain terminologies when a user specifies a use case specification with the RUCM editor, and dynamically constructing the dictionary containing identified domain terminologies and their domain classifications. When a user completes her/his work, RUCMBot automatically updates the dictionary by re-clustering and extracting terms, and subsequently realises the self-updating of the dictionary.

3. Features of RUCMBot Tool

1. Domain Classifier

RUCMBot can achieve real-time domain classification based on users input. the index =0 means the use case specification belongs to the domain marked 0 (the labels will be reassigned after re-clustering).

2. Domain Terminology Recommendation

RUCMBot can recommend domain terms based on prefixes and the predicted POS of a sentence after the domain classification is successful.

3. Domain Terminology Dictionary Auto-Construction

This is part of the domain terminologies dictionary, including Domain 0 (Home Automation), Domain 1 (Energy Production) and Domain 2 (Autonomous Driving). Along with the continuous use of RUCM for use case modeling, the dictionary will be self-updating.