Papers by Hsien-Tsung Chang
Computational Intelligence and Neuroscience, 2016
Leisure travel has become a topic of great interest to Taiwanese residents in recent years. Most ... more Leisure travel has become a topic of great interest to Taiwanese residents in recent years. Most residents expect to be able to relax on a vacation during the holidays; however, the complicated procedure of travel itinerary planning is often discouraging and leads them to abandon the idea of traveling. In this paper, we design an automatic travel itinerary planning system for the domestic area (ATIPS) using an algorithm to automatically plan a domestic travel itinerary based on user intentions that allows users to minimize the process of trip planning. Simply by entering the travel time, the departure point, and the destination location, the system can automatically generate a travel itinerary. According to the results of the experiments, 70% of users were satisfied with the result of our system, and 82% of users were satisfied with the automatic user preference learning mechanism of ATIPS. Our algorithm also provides a framework for substituting modules or weights and offers a new method for travel planning.
A Switching Proxy for Web Search Engines
Advances in information sciences and services (AICIT), Jun 30, 2011
A framework of hybrid methods for splog detection
ABSTRACT More and more people setup their own blogs due to the popularization of web 2.0 and matu... more ABSTRACT More and more people setup their own blogs due to the popularization of web 2.0 and maturation of Internet techniques. People love to share their thoughts and moods even new products experiences through their blogs. This phenomenon pushes the business models and web technologies forward in a rapid speed. It is also becoming an important research topic of how to improve the search results by using splog detection method. In this paper, we proposed a framework combining five hybrid methods to detect the splogs. In our experiment shows that there is no single method can optimally detect all splogs. We use simple and fast methods to detect the splog to decrease the number of remaining blog for the complex methods. It will successfully decrease the time needed in detecting the splogs. The results of the detection can feedback to the framework, and improve the overall detection performance.
Gerontechnology, Jun 3, 2014
the formation and cooperation of interdisciplinary teams in knowledge integration 2 and innovativ... more the formation and cooperation of interdisciplinary teams in knowledge integration 2 and innovative R&D processes from 'creative idea' to 'feasibility assessment' to 'prototyping the product'. The R&D process has contributed to a new innovation model, which will also be introduced in the symposium. In addition, some gerontechnology products developed in this program will be introduced by the presenters, such as an interactive system, health management system, intelligent transport, and rehabilitation game for the elderly.
Annual Conference on Computers, Jul 23, 2009
Due to the rapid growth of World Wide Web and the popularization of personal computer, the number... more Due to the rapid growth of World Wide Web and the popularization of personal computer, the number of World Wide Web users is increasing in an incredible speed. It makes the need of disk space for a large web service also increases in the same rate. Therefore, it becomes to an important technique to design and implement a powerful network file system for the large web service providers. In this paper we proposed three design issues for the scalable network file system. We use the variable number of objects within a bucket to decrease the internal fragmentation in small files. We also proposed the free space and access load-balancing mechanism to balance the overall loading within the bucket servers. And last we proposed a mechanism for caching data which is frequently accessed to lower the total number of disk I/O. These proposed mechanisms will effectively improve the performance of the scalable network file system for the large web services.
Lecture Notes in Computer Science, 2013
In recent years, more and more researchers try to make Microsoft Kinect and Augmented Reality (AR... more In recent years, more and more researchers try to make Microsoft Kinect and Augmented Reality (AR) into real lives. In this paper, we try to utilize both Kinect and AR to build a dynamic fitting room. We can automatically measure the clothes size of a user in popular brands or different country standards. A user can utilize gesture to select cloths for fitting. Our proposed system will project the video dynamically of dressing selected clothes in accordance with the captured video from Kinect. This system can be utilized in clothing store, e-commerce of clothes shopping, and at your home when you are confusing choosing a clothes to wear. This can greatly reduce the time you fitting clothes.
A Topic Detection and Tracking System with TF-Density
Lecture notes in electrical engineering, 2013
In the past, news consumption took place predominantly via newspapers and were hard to track. Now... more In the past, news consumption took place predominantly via newspapers and were hard to track. Nowadays, the rapid growth of the Internet means that news are continually being shared and stored on a previously unimaginable scale. It is now possible to access several news stories on the same topic on a single web page. In this paper, we proposed a topic detection and tracking system with a new word measurement scheme named TF-Density. TF-Density is a new algorithm modified from the well-known TF-IWF and TF-IDF algorithms to provide a more precise and efficient method to recognize the important words in the text. Through our experiments, we demonstrated that our proposed topic detection and tracking system is capable of providing more precise and convenient result for the tracking of news by users.
Touch-Panel Usability for Children
In the past few years, touch panels have increasingly been used in children’s education, and more... more In the past few years, touch panels have increasingly been used in children’s education, and more research has been devoted to the development of electronic schoolbag. However, children often encounter difficulties operating touch screens, possibly due to their relative lack of experience using computers or touch panels, or because of the small size of their hands. This study investigates the drag, zoom and rotation operations of 4.3-inch and 10.1-inch touch panels, aiming to provide recommendations for appropriate movement speed settings. The operating gestures used by child users were also observed. The movement rate of 30 young adults using 4.3-inch and 10 inch touch panels were taken as base values, and compared to the movement rate of 30 other children serving as test subjects. Overall, children performed better on the 10.1-inch panel than on the 4.3-inch panel, thus 10.1-inch panels are recommended for use in electronic schoolbag for children. In addition, results for the chil...
PLOS ONE, Nov 12, 2019
This study presents a novel research approach to predict user interaction for social media post u... more This study presents a novel research approach to predict user interaction for social media post using machine learning algorithms. The posts are converted to vector form using word2vec and doc2vec model. These two methods are used to analyse the best approach for generating word embeddings. The generated word embeddings of post combined with other attributes like post published time, type of post and total interactions are used to train machine learning algorithms. Deep neural network (DNN), Extreme Learning Machine (ELM) and Long Short-Term Memory (LSTM) are used to compare the prediction of total interaction for a particular post. For word2vec, the word vectors are created using both continuous bag-of-words (CBOW) and skip-gram models. Also the pre-trained word vectors provided by google is used for the analysis. For doc2vec, the word embeddings are created using both the Distributed Memory model of Paragraph Vectors (PV-DM) and Distributed Bag of Words model of Paragraph Vectors (PV-DBOW). A word embedding is also created using PV-DBOW combined with skip-gram.
Nowadays, Traffic accidents (TAs) causes much damage regarding human and asset. TAs records have ... more Nowadays, Traffic accidents (TAs) causes much damage regarding human and asset. TAs records have been stored and published in many Open Data sources. In this article, we propose a method of using clustering algorithm Densitybased Spatial Clustering of Applications with Noise (DBSCAN) to classify the TAs' records to find the Density Traffic Accident Areas (DTAA). We also discuss the optimal variables in DBSCAN that need to consider when applied in real urban areas. We emphasize the characteristics of DTAA by the Bound Value (BV) and modeling some important traffic characteristics. We then evaluate the performance and efficiency between DBSCAN and K-mean clustering methods. The result clusters and characteristics can be easily adapted to real traffic applications for increase the travel safety.
HieraRoute: A Dynamic and Hierarchical Road Routing Algorithm Based on Real-Time Information
Journal of Internet Technology, Sep 1, 2014
ABSTRACT As mobile computing technology undergoes rapid advancement, automobile navigation system... more ABSTRACT As mobile computing technology undergoes rapid advancement, automobile navigation system costs are plummeting and are no longer considered expensive luxuries. Indeed, many people now use these gadgets on a daily basis and carry them constantly. However, the computing power of car navigation device is usually limited. In this paper, we utilize the hierarchical concept dividing the road information into two levels to reduce the overall computation complexity; the Low-Level Graph adopts each intersection as a node and each road connecting two nodes as an edge. The node used in the High-Level Graph is called a PopNode (Popular Node), which refers to the most frequently used in low-level node. Our proposed algorithm adopts the two levels information to calculate the route and proof that our algorithm can reduce the computational complexity but also generate the route that is closed to the optimal result. We also imply random function and dynamic concepts into the algorithm to improve the driving experience.
Is It Possible to Use Chatbot for the Chinese Word Segmentation?
A word is the smallest item in Natural Language Processing. However, there is no obvious boundary... more A word is the smallest item in Natural Language Processing. However, there is no obvious boundary for Chinese words. How to segment Chinese words always obstructs Chinese researches and applications. Nowadays, a neural network model, Seq2Seq with LSTM, is well-known for translation or chatbot application. In this paper, we try to transform the Chinese word segmentation problem into a translation problem. And we utilized an open-source chatbot to simulate the translation task. In our experimental results, we can produce similar Chinese word segmentation results when we provide training data which is automatically generated from famous Chinese word segmentation services.
WSEAS Transactions on Information Science and Applications archive, Aug 1, 2008
Due to the popularity of digital cameras and web authors'enriching the visual aesthetics, the num... more Due to the popularity of digital cameras and web authors'enriching the visual aesthetics, the number of web images is growing in an uncontrolled speed. The images in the World Wide Web are becoming a large image library for browsing. It is an important issue that how to retrieve the images accurately on the World Wide Web. In this paper we describe the architecture of the web image retrieval systems with automatic image annotation techniques. And we propose four methods to generate the annotation automatically for every image from its hosted web page, by analyzing the structural blocks, collecting anchor text of link structures, and gathering shared annotation with other images with the same visual signature.
An e-caring chair for physiological signal measurement and recording
Medical Engineering & Physics, Feb 1, 2013
There is an increasing awareness among the populace of the need for regular health check-up to de... more There is an increasing awareness among the populace of the need for regular health check-up to detect diseases in their early stages and thereby administer treatments in a timely fashion. However, commercially available physiological signal monitoring devices, which may offer clues on the onset of diseases, are time-consuming, far from user friendly and limited in their applications. We design an e-caring chair that combines six modular physiological signal measurement instruments into a single unit, enabling users to simultaneously measure the blood pressure, body temperature, heart rate, height, weight and body fat percentage, and display the results and simple diagnoses in real time. The e-caring chair further allows for easy integration of additional physiological signal measuring devices, speedy measurements and long term monitoring of any trends that may emerge, making it easier for users to be alerted to physiological changes in the body without the need to enlist assistance from medical personnel. In this paper, we describe how this e-caring chair can be placed in several different environments for different purposes.
Is It Possible to Use Chatbot for the Chinese Word Segmentation?
Proceedings of the 2019 3rd International Conference on Natural Language Processing and Information Retrieval
A word is the smallest item in Natural Language Processing. However, there is no obvious boundary... more A word is the smallest item in Natural Language Processing. However, there is no obvious boundary for Chinese words. How to segment Chinese words always obstructs Chinese researches and applications. Nowadays, a neural network model, Seq2Seq with LSTM, is well-known for translation or chatbot application. In this paper, we try to transform the Chinese word segmentation problem into a translation problem. And we utilized an open-source chatbot to simulate the translation task. In our experimental results, we can produce similar Chinese word segmentation results when we provide training data which is automatically generated from famous Chinese word segmentation services.
An Efficient Multilevel Healthy Cloud System Using Spark for Smart Clothes
2016 International Computer Symposium (ICS)
Nowadays, we can find part of people into the suboptimal health status. However, most of them hav... more Nowadays, we can find part of people into the suboptimal health status. However, most of them have not enough time for health examinations regularly. Smart clothes which are wearable and convenient devices for monitoring physiological signals solve this problem. In this study, we develop an efficient multilevel healthy cloud system to analyzing data collected from CGU smart clothes. The cloud system classifies the raw data into three types: normal data, continuous data, and multimedia data. After that, the cloud system analyzes data and predict diseases using Apache Spark. And then, the system will transfer final results to smartphones. The experimental result shows that the cloud system achieves very high-speed performance. Moreover, the system has friendly health information presentation and effective authorization process to help doctors and patients contact each other.
PeerJ Computer Science
Urdu is a widely used language in South Asia and worldwide. While there are similar datasets avai... more Urdu is a widely used language in South Asia and worldwide. While there are similar datasets available in English, we created the first multi-label emotion dataset consisting of 6,043 tweets and six basic emotions in the Urdu Nastalíq script. A multi-label (ML) classification approach was adopted to detect emotions from Urdu. The morphological and syntactic structure of Urdu makes it a challenging problem for multi-label emotion detection. In this paper, we build a set of baseline classifiers such as machine learning algorithms (Random forest (RF), Decision tree (J48), Sequential minimal optimization (SMO), AdaBoostM1, and Bagging), deep-learning algorithms (Convolutional Neural Networks (1D-CNN), Long short-term memory (LSTM), and LSTM with CNN features) and transformer-based baseline (BERT). We used a combination of text representations: stylometric-based features, pre-trained word embedding, word-based n-grams, and character-based n-grams. The paper highlights the annotation guid...
A Topic of Interest-based Approach on Online Social Network Services for Information Sharing
Proceedings of the Fifth International Conference on Network, Communication and Computing, 2016
Updating status messages on social network become one of the main activities of the Internet user... more Updating status messages on social network become one of the main activities of the Internet users. However, sharing a story become more useful when it can be addressed appropriately to other users. In this article, we propose a novel approach to sharing the status message appropriately, to enhance the usefulness of the information according to user interest. We emphasize to leverage state-of-the-art topic modeling technique and similarity measurement to obtain topic interest and similarity between users respectively. We evaluated our approach using data gathered from real-world Twitter datasets and English Wikipedia. In the end, the proposed approach can find the users appropriately on their interest in English tweets. We conclude that this approach can be adopted in different social networks platform for sharing the information to specific user or group for the usefulness of information to the social networks users.
The Proceedings of the 5th International Conference on Industrial Application Engineering 2017, 2017
Nowadays, Traffic accidents (TAs) causes much damage regarding human and asset. TAs records have ... more Nowadays, Traffic accidents (TAs) causes much damage regarding human and asset. TAs records have been stored and published in many Open Data sources. In this article, we propose a method of using clustering algorithm Densitybased Spatial Clustering of Applications with Noise (DBSCAN) to classify the TAs' records to find the Density Traffic Accident Areas (DTAA). We also discuss the optimal variables in DBSCAN that need to consider when applied in real urban areas. We emphasize the characteristics of DTAA by the Bound Value (BV) and modeling some important traffic characteristics. We then evaluate the performance and efficiency between DBSCAN and K-mean clustering methods. The result clusters and characteristics can be easily adapted to real traffic applications for increase the travel safety.
HieraRoute: A Dynamic and Hierarchical Road Routing Algorithm Based on Real-Time Information
Journal of Internet Technology, 2014
As mobile computing technology undergoes rapid advancement, automobile navigation system costs ar... more As mobile computing technology undergoes rapid advancement, automobile navigation system costs are plummeting and are no longer considered expensive luxuries. Indeed, many people now use these gadgets on a daily basis and carry them constantly. However, the computing power of car navigation device is usually limited. In this paper, we utilize the hierarchical concept dividing the road information into two levels to reduce the overall computation complexity; the Low-Level Graph adopts each intersection as a node and each road connecting two nodes as an edge. The node used in the High-Level Graph is called a PopNode (Popular Node), which refers to the most frequently used in low-level node. Our proposed algorithm adopts the two levels information to calculate the route and proof that our algorithm can reduce the computational complexity but also generate the route that is closed to the optimal result. We also imply random function and dynamic concepts into the algorithm to improve the...
Uploads
Papers by Hsien-Tsung Chang