(1) Yu Sheng - Customizable Poem Writing System
This project provides a classical Chinese poem writing system with the broadest coverage among all existing online systems. "Yu Sheng" proposes a 3-degree Human-Computer Interactive poem writing strategy for different users' need.
Ma Jing Kun
(2) UMor-PG - The Intelligent Chatbot for Postgraduate Admission Information!
Yao Yutong
Can You Read Cantonese? Candarin Can Help!
Cantonese is a common Chinese dialect. Also, it is a low resource language for natural language processing. Inadequacy resources are hard to support neural machine learning tasks. Our works focus on finding the solutions in neural machine translation. We will show our methodologies. Then, to make our works more expressive, we take web technology to combine our project.
Chong Iok Hei and Ku Su Wa
UMor: A UM Chatbot!
UMor is an ionic Chatbot of University of Macau (UM). Besides chatting, UMor has been designed to answer questions related to UM, acting as an intelligent assistant to explore various academic activities and informatoin regarding the programs and facilities, residential colleges, restaurants, as well as the in-campus transportation status. Umor makes use of the most cutting-edge technologies of natural language processing and deep learning models to deal with the questions expressed in natural language. In order to deal with dialogues with paraphrasing, UMor analyzes the questions by modeling the semantic meaning of the sentences instead of its surface forms, making it more robust. UMor has been deployed as a mini program on the popular WeChat platform. The current UMor also provides simple games and the daily horoscopes fo rthe entertainment purpose.
LI Junfei and CHEN Guanhua
iLearner@2019: An Intelligent e-Learning Platform for Teaching & Learning
Online electronic learning (i.e., e-Learning) systems have been popular in recent years, especially they have been widely used by students in learning different subjects at the primary and secondary school. However, e-learning platforms on Portuguese learning are quite rare. iLearner aims at providing an online system for Portuguese students in the university, focuses on learning Portuguese verb conjugations. Besides, the system can automatically generate questions and correct the tests automatically, which can reduce the workload of teachers. In addition, we provide a superior search engine for teachers searching existing questions to build up questions set on the fly. With the system, the students can study anytime, anywhere as long as they have connection to the Internet.
Kuok Chon Kit and Wong Kuan Hou
Automatic Generation of Test Questions for Language Learning
There are many kinds of English online learning websites, while almost none of them provides support of Portuguese. So we made a site for Portuguese learners, which provides services for both learners and teachers. This project provides the functions of automatic question generation for teachers. With the help of Analysis program, sentences will be analyzed and parsed. The verb in the sentence will be picked up, and the candidates that have the different tenses of the verb will be generated. It makes this sentence becomes a multiple choice question. This project also provides the user interface for teachers and students to manage exercises and do tests.
Yang Yuhan
(1) UM-iMethod2 Chinese-Portuguese Translation Oriented Interactive Input Method (IME) System
Um-iMethod is an advanced MT system integrated with the SMT mechanism. In recent, it only provides Chinese to English translation service and still has some shortages. Nowadays, enormous Chinese-Portuguese translation demands appear in Macau since Macau acts as an important bridge between Portuguese-speaking countries and China. UM-iMethod2 is the upgraded version of UM-iMethod to provide an efficient way to produce Chinese-English, or Chinese-Portuguese translation works.
Tong Weng Hou, Felix
(2) Neural Machine Translation Systems
Neural Machine Translation (NMT) is a recent and hot proposed approach in machine translation area. Unlike the SMT, the NMT focuses on building a neural network that can maximize the translation performance by tuning the parameters of the network. Typical NMT system is built within the RNN based encoder-decoder framework. In this framework, the encoder RNN Processes the words in a source language sentence into fixed length vectors, which is as the inputs to decoder RNN to decode the translation sentence. This project is based on an existing open source NMT framework to modify and build a server which can replace the SMT server without changing any code of the modern Machine Translation business framework, and an aligner system which can automatically extract rare corpus from comparable corpus such as Wikipedia.
Hou Tim LAI, Revo
(1) UM-iMethod Statistical Machine Translation (SMT) Oriented Input Method Editor
Um-iMethod is a web-based input method editor integrated with Statistical Machine Translation. It reuses information produced during SMT decoding process. It helps translators to work more efficiently and bring machine translation to users in a different way.
Pui Lam Si, Bryce
Recognition of Numerical Expression in Parallel Corpus in SMT
In this project, we realize a Chinese to English statistical machine translation system using the Moses decoder, which is a part of a statistical machine translation toolkit. In the current model, the translation quality is not satisfactory. One of the reasons is the poor translation of the numerical expressions embedded in the text to be translated. The main objective of this project is to recognize those expressions and generalize it by replacing them with placeholders. It aims to reduce the interference of the numerical values during the translation, and hence to improve the performance of the system.
Li Xun
iFinger - Study of Gesture Recognition Technologies & Its Applications
This system can extract the hand features from images using common digital camera. So that we can calculate the human hands shape, motion, moving direction, moving speed, etc. In addition, we added some anti-shaking algorithms to stable the result. Hence, it can control computer to do some tasks only with human hands and show the result in real time.
Ao Chi Hong (Ben) and Choi Chi Ian (Anika)
(1) iINTERACT - Study of Eye Tracking Technologies and Its Applications
iINTERACT, a video-based Eye-tracking system that monitors and captures the details of the fast movement of subject’s pupil. The system employs an inexpensive solution by using convention digital video camera to keep track of eye’s movement and control for human-computer interaction. It allows users to play computer games, input text and draw pictures with eye control instead of keyboard and mouse. In addition, the gaze motion can be analyzed and displayed in form of gaze plot and heat map.
Lei I Kei (Wendy) and Wong Wong (Jacky)
(2) CanSpeak: Cantonese-to-English Real-Time Speech Translation System
CanSpeak is a Cantonese Text to Speech application over Web environment. The system accepts Traditional Chinese text input and produces corresponding Cantonese pronunciations. User can easily rewind and listen to the synthesized voices.
Lei Sio Meng (Vincent), Tang Kam Wa (Roy) and Ng Ka Long (Ronald)
(3) Cyclops II: Snapshot Mobile Translation System Based on Thin Client-Server Architecture
Smartphone and wireless network is popular today. People can use their device to access Internet easily. Therefore, we have an idea to make an application to do snapshot translation via smartphone and Internet. The advantage of this application is the user can do the translation without input the text. This project is an extension of Cyclops, the mobile-based snapshot translation. The recognition of text and machine translation is carried out at a backend server, through the use of mobile device.
Wong Cheok Meng (Jack) and Chan In Wa (Edward)
"KERST!" is a Cantonese-to-English Real-Time Speech Translation System.
The system aims to provide a convenient translation system for users who speak Cantonese and would like to translate their spoken words into English. The system targeted for users who can speak Cantonese and especially those users who cannot speak English and this will be a great tool for them to communicate to others who speak English.KERST comprises three sub-systems: Cantonese Speech Recognition System, Chinese-to-English Text Translation System and English Text-to-Speech Conversion System. The user just needs to speak to the system using microphone and the corresponding English speech will be read out.
Lei Ka Man(Kobi), Lao Hou Wang(Van) and Yung Weng Hong
"iTagIt!" Automatic Image Tagging System
"iTagIT!" is a automatic image tagging system. Users selects an image they would like to tag and input some parameters to help software segment the image, then it extracts predefined features from the segmented regions and uses the pre-trained classifier to assign keyword to the regions. Then users can manage their pictures based on these keywords.
Zhu Hongyuan (Ballman, D-A6-2857-2) and Cui Chi (D-A6-2870-6)
(1)Compact OCR Based on Peripheral Direction Contributivity and Its Application in Pocket Based Snapshot(Chinese) Translation(Cyclops)
Cyclops system allows users to capture a photo of Chinese characters as the source input of the translation. The system gets the features of the input Chinese characters based on Peripheral Direction Contributivity (PDC). It transforms the features into a vector by using K-L transform, and searches in the features database to find out the most suitable one by calculating the Euclidean distance. Finally, it translates the Chinese into English and Portuguese.
Chan Wai Kit (Sam, D-A5-2721-5), Leong San Weng (Brandon, D-A5-2714-3)
(2)Translation Knowledge Sharing System, Based on Liferay Environment
This project is a system combining two influential topics (Machine translation and Web 2.0), and our objective is to provide a web based platform for users to translate and share their translation knowledge. Although we are not dedicated to implement the translation algorithm, our goal is to bring the existed system, PCT to an online platform.
Ma Cheng Man (Emmy), Au Ka Heng (Kyle)
English to Chinese Machine Translation based on Constraint Synchronous Grammar (CSG)
English to Chinese Translation System based on Constraint Synchronous Grammar catches the meaning of the English sentence, then performs the meaning in Chinese. It will not be a word by word translation system because we are using a method based on linguistic rules, which means that words will be translated in a linguistic way - the most suitable (orally speaking) words of the target language will replace the ones in the source language. The system performs translation through the user-friendly interface in the most popular office application - Microsoft Word, it also provides mouse tracking translation function which gives the most convenient to users.
Hong Iok Sai (Su), Chio Kin Tong (Tony), Choi Wa Seong (Nelson), Wong Kuok Chu (Barry)
(1)Pak Pak serves you - Application on Interactive Dialogue and Answering (IDQA) System
Leong Pok Man (Luis), Pun Ka I (Violet), Mak Hoi Fong (Leo)
(2)Netrans - Network based Translation System
Netrans provides an Intranet based translation environment, in order to provide knowledge sharing. Multiple translation modes are provided for document translation. Users can share and reuse the translation resources in a working space.
Hao Cheong Cheong, Leung Kwok Kit, Tou Hoi Wai