A B C D E F G H I J K L M N O P Q R S T U V W X Y Z All
H. Mustafa, Haidy
- Automatic Requirement Classification Technique: Using Different Stemming Algorithms
Authors
Source
Data Mining and Knowledge Engineering, Vol 10, No 6 (2018), Pagination: 122-127Abstract
Requirement Engineering is the first crucial stage in the software life cycle. Classifying those requirements into functional and non-functional requirements is an important activity during requirement engineering process. As a result of requirement engineering process, a software requirement specification document is produced. This document contains a detailed description of all requirements written using natural language. The automatic processing of natural language is not an easy task. Since natural language is full of ambiguity, has no formal structure, and very variable. This paper presents an automatic classification of requirements into functional and non-functional requirements using two machine learning algorithms. In this paper, different stemming techniques are used to address some of natural language challenges. A dataset of 625 requirements (functional and non-functional) is used to train and test the machine learning model. The experiments showed that some stemming techniques increased the performance than other stemming techniques.
Keywords
Requirement Classification, Non-Functional Requirements, Stemming, Software Projects, Functional Requirements.
- Supervised Technique for Arabic Sentiment Analysis Using Different Preprocessing Methods and Features
Authors
Source
Data Mining and Knowledge Engineering, Vol 10, No 8 (2018), Pagination: 160-166Abstract
The widespread of social media websites resulted to produce a massive amount of data every single minute. This kind of data represents people’s opinions, attitudes and feedback about different topics, political decisions, and products. Processing and analyzing such kind of data in order to understand people’s thoughts, feedback, and needs is called sentiment analysis (opinion mining). Sentiment analysis becomes a hot area nowadays because of this rapid growth in social media websites. Sentiment analysis main task is classifying text/documents/words according to its polarity into positive/negative opinions. The traditional sentiment analysis techniques are supervised, unsupervised and semi-supervised. Theses traditional techniques still didn’t give a high quality while working with Arabic language. Due to the complexity of Arabic language and its high derivatives, the sentiment analysis task becomes complicated and not an easy task. This paper applies different supervised techniques (machine learning algorithms) with different preprocessing methods in order to investigate their importance.