Browsing by Author "Islam, M.U.,"

Filter results by typing the first few letters
Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Polarity detection of online news articles based on sentence structure and dynamic dictionary
    (Institute of Electrical and Electronics Engineers Inc., 2017-07-02) Islam, M.U.,; Ashraf, F.B.,; Abir, A.I.,; Mottalib, M.A.
    The importance of online news article has evolved notably with the advancement of information and technology. However, some of the news are violent as well as obnoxious. So, identifying and categorizing online news article automatically is important as well as remains challenging. Using opinion mining and sentiment analysis, we propose an intuitive approach of detecting positive or negative news from an online news article. Our approach consists of a sentence identification phase, followed by a dynamic library of predefined negative and positive strings and at last marking whether the paragraph is positive, negative or neutral. Our approach detects the polarity of online news articles with around 91% accuracy rate. Sentence type identification before using dynamic dictionary of positive and negative words is the key factor which resolves the issue of finding out the part of the sentence which holds the polarity of the sentence.
  • Thumbnail Image
    Item
    Yoga posture recognition by detecting human joint points in real time using microsoft kinect
    (Institute of Electrical and Electronics Engineers Inc., 2018-02-09) Islam, M.U.,; Mahmud, H.,; Bin Ashraf, F.,; Hossain, I.,; Hasan, M.K.
    Musculoskeletal disorder is increasing in humans due to accidents or aging which is a great concern for future world. Physical exercises can reduce this disorder. Yoga is a great medium of physical exercise. For doing yoga a trainer is important who can monitor the perfectness of different yoga poses. In this paper, we have proposed a system which can monitor human body parts movement and monitor the accuracy of different yoga poses which aids the user to practice yoga. We have used Microsoft Kinect to detect different joint points of human body in real time and from those joint points we calculate various angles to measure the accuracy of a certain yoga poses for a user. Our proposed system can successfully recognize different yoga poses in real time.